Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Strategies to reduce transportation emissions in India : identifying air quality and climate co-benefits… Reynolds, Conor Charles OBrien 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2011_spring_reynolds_conor.pdf [ 23.23MB ]
Metadata
JSON: 24-1.0071479.json
JSON-LD: 24-1.0071479-ld.json
RDF/XML (Pretty): 24-1.0071479-rdf.xml
RDF/JSON: 24-1.0071479-rdf.json
Turtle: 24-1.0071479-turtle.txt
N-Triples: 24-1.0071479-rdf-ntriples.txt
Original Record: 24-1.0071479-source.json
Full Text
24-1.0071479-fulltext.txt
Citation
24-1.0071479.ris

Full Text

STRATEGIES TO REDUCE TRANSPORTATION EMISSIONS IN INDIA: IDENTIFYING AIR QUALITY AND CLIMATE CO-BENEFITS FOR THE DEVELOPING WORLD  by  CONOR CHARLES OBRIEN REYNOLDS B.A., B.A.I., University of Dublin, Trinity College, 1998 M.A.Sc., University of British Columbia, 2002  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Resource Management and Environmental Studies) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) November 2010  ! Conor Charles OBrien Reynolds, 2010  ABSTRACT  Emissions from on-road transportation sources are a complex mixture of gaseous and particulate pollutants. Particulate matter (PM) emissions are especially important because – although short lived in the atmosphere – they are strongly associated with cardiovascular and respiratory disease and are strong climate forcing agents. The overall objective of this research was to quantify the effectiveness of emission control policies for in-use vehicles in India. I have focused on understanding the impacts of large-scale adoption of compressed natural gas (CNG) as an alternative to diesel and gasoline in New Delhi, India. In Chapter 2, I quantified the climate impacts of switching to CNG for public transportation vehicles (taxis and buses). The study showed that converting buses from diesel to CNG significantly reduced climate-warming diesel particulate matter (PM), but the increase in CH4 emissions from all vehicle types offset much of this benefit. Chapters 3-5 focused on auto-rickshaws (three-wheeled taxis), which are an important mode of passenger transport in many developing countries. In Chapter 3, a survey of 350 drivers quantified activity patterns, fuel consumption and CO2 emissions for auto-rickshaws, and a model was developed to better understand the determinants of visible smoke emissions. Chapter 4 describes a laboratory (chassis dynamometer) study that measured emissions from 31 auto-rickshaws, and establishes fuel-based emission factors for gaseous and fine PM pollutants from 2-stroke and 4-stroke spark-ignited engines fueled with CNG and gasoline. Finally, Chapter 5 examines a range of emission-reduction policies for auto-rickshaws, including phasing out 2stroke engines, switching to CNG fuel, scrapping older vehicles and four different types of inspection and maintenance (I/M) programs. Together, these studies demonstrate that certain fuel/engine combinations, such as CNG-fueled 4-stroke engines, are more robust low-emitters than others, and can be an effective alternative to diesel engines (in buses) or 2-stroke engines (in auto-rickshaws). Although this research has examined emissions-reduction policy in New Delhi, the findings are applicable to in-use vehicles in many other jurisdictions in the developing world.  ii  PREFACE Each research chapter of this dissertation (Chapters 2-5) was written as a stand-alone manuscript for publication in the peer-reviewed academic literature, and parts of Chapter 1 are based on a co-authored book chapter. In the following I provide details of my contributions to the literature reviews, research activities, data analysis and manuscript preparation for the co-authored publications that have been produced during my doctoral studies. Chapter 1: General introduction A portion of this introductory chapter is based on a co-authored book chapter (Sections 1.3-1.5). I conducted the literature review and wrote the manuscript for the book chapter, and Andrew Grieshop and Milind Kandlikar contributed revisions and editorial input. A version of Chapter 1, Sections 1.3-1.5, has been published: Reynolds, C. C. O., Grieshop, A. P. and Kandlikar, M. Reducing Particulate Matter Emissions from Buses and Trucks in Asia: A Framework to Assess Air Pollution and Climate Change Co-Impacts, in Zusman, E., Srinivasan, A. and Dhakal, S. (eds) Low Carbon Transport in Asia: Strategies for Optimizing Co-benefits, Earthscan, London, (in press) Chapter 2: Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in Delhi For this paper, I conducted the literature review, designed and developed the analytical framework, conducted the policy analysis, and wrote the manuscript. Milind Kandlikar contributed to the conceptual development of the study, provided substantial guidance throughout the analysis, and assisted with revision of the manuscript. A version of chapter 2 has been published: Reynolds, C. C. O.; Kandlikar, M. Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860-5865. Chapter 3: Key determinants of PM and GHG emissions from natural gas fueled auto-rickshaws in Delhi For this study, I designed the research, prepared the survey instrument, pilot-tested the survey, and then trained 5 Indian research assistants to survey auto-rickshaw drivers in Delhi. I analyzed the data and wrote the manuscript. Madhav Badami was instrumental during the conception of the study and development of the survey instrument, and he and Milind Kandlikar provided comments for revision of the manuscript. A version of chapter 3 has been provisionally accepted for publication (pending revisions): Reynolds, C. C. O.; Kandlikar, M.; Badami, M. G. Key determinants of PM and GHG emissions from natural gas fueled auto-rickshaws in Delhi. Transp. Res. D–Transp. Environ. iii  Research ethics approval was obtained for the survey of auto-rickshaw drivers in Delhi. The UBC Behavioural Research Ethics Board (BREB) designated the research ‘Minimal Risk’, and Certificate Number H08-01517 was awarded on August 6, 2008. Renewal of the certificate was obtained on June 24, 2009. Chapter 4: Climate and health relevant emissions from in-use Indian three-wheelers fueled by natural gas and gasoline Milind Kandlikar and I came up with the idea for this study. I negotiated the contractual terms for conducting the research in an Indian vehicle-testing laboratory (the International Center for Automotive Technology at Manesar, near Delhi). Andrew Grieshop and I worked together to design the research protocol, prepare specialized equipment for data collection, arrange recruitment of test-vehicles, and conduct the research in India. Both Milind Kandlikar and Andrew Grieshop provided critical input at the data analysis stage, and assisted with interpreting the results. I wrote the manuscript, and Andrew Grieshop and Milind Kandlikar provided feedback for revisions. A version of chapter 4 has been submitted: Reynolds, C. C. O.; Grieshop, A. P.; Kandlikar, M. Climate and health relevant emissions from in-use Indian three-wheelers fueled by natural gas and gasoline. Chapter 5: Fuels, technology and vehicle maintenance: Assessing strategies to reduce emissions from Indian auto-rickshaws For this paper, I conceived of the study idea, built the analytical model (using primary data collected during the research conducted for Chapter 5, as well as other data obtained from a literature review) and wrote the manuscript. Andrew Grieshop and Milind Kandlikar contributed valuable discussions about the analysis, and gave feedback that helped me to improve the manuscript. A version of chapter 5 will be submitted: Reynolds, C. C. O.; Grieshop, A. P.; Kandlikar, M. Fuels, technology and vehicle maintenance: Assessing strategies to reduce emissions from Indian auto-rickshaws.  iv  TABLE OF CONTENTS  ABSTRACT ............................................................................................................................... ii! PREFACE ................................................................................................................................. iii! TABLE OF CONTENTS ............................................................................................................v! LIST OF TABLES .................................................................................................................. viii! LIST OF FIGURES.....................................................................................................................x! LIST OF ABBREVIATIONS .................................................................................................. xii! ACKNOWLEDGEMENTS ......................................................................................................xv! DEDICATION ....................................................................................................................... xvii! 1.! Chapter 1: General Introduction..........................................................................................1! 1.1.! Context: Air pollution and emissions from vehicles in India............................................1! 1.1.1.! Motivation ..................................................................................................................1! 1.1.2.! Why study three-wheeled auto-rickshaws in Delhi? ..................................................2! 1.1.3.! Links between air quality and climate change ...........................................................4! 1.1.4.! Research scope and strategy ......................................................................................5! 1.2.! Research objectives ...........................................................................................................7! 1.3.! Transportation energy use and its impacts ........................................................................8! 1.3.1.! Transportation activity and fuel-use in India.............................................................8! 1.3.2.! Pollutant emissions from vehicle engines ................................................................10! 1.3.3.! Air quality and health effects of traffic-related air pollution...................................12! 1.3.4.! Climate impacts of traffic-related air pollution .......................................................13! 1.3.5.! Quantifying emissions from mobile sources: Emission inventories.........................15! 1.4.! Emission control policies in Asia....................................................................................17! 1.4.1.! Emissions standards for new vehicles ......................................................................18! 1.4.2.! Exhaust aftertreatment devices for new and in-use vehicles....................................20! 1.4.3.! Fuel quality regulations ...........................................................................................21! 1.4.4.! Emission control policies for the in-use fleet ...........................................................22! 1.4.5.! Alternative fuels........................................................................................................24! 1.4.6.! New powertrain technologies...................................................................................26! 1.5.! Framework to evaluate co-impacts..................................................................................27! 1.5.1.! Addressing critical knowledge gaps.........................................................................30! 1.6.! Overview of dissertation .................................................................................................31! 1.7.! References .......................................................................................................................34! 2.! Chapter 2: Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in Delhi .......................................................................................43! 2.1.! Introduction .....................................................................................................................43! 2.2.! Methods ...........................................................................................................................44! 2.2.1.! Fuel efficiency and emissions factors.......................................................................44! 2.2.2.! Global warming/cooling metrics for climate forcing aerosols ................................47! 2.3.! Results .............................................................................................................................50! 2.3.1.! Change in CO2-equivalent emissions .......................................................................50! 2.3.2.! Uncertainty in emissions factors ..............................................................................52! 2.3.3.! Uncertainty in GWP/GCP ........................................................................................54! v  2.4.! Discussion .......................................................................................................................55! 2.5.! References .......................................................................................................................58! 3.! Chapter 3: Determinants of PM and GHG emissions from natural gas fueled autorickshaws in Delhi .......................................................................................................................61! 3.1.! Introduction .....................................................................................................................61! 3.2.! Methods ...........................................................................................................................62! 3.2.1.! Structured survey......................................................................................................62! 3.2.2.! Calibration of PM observations ...............................................................................63! 3.2.3.! Determining correlation between survey variables and ‘high-PM emitters’...........64! 3.3.! Results .............................................................................................................................65! 3.3.1.! Fleet characteristics, vehicle activity and GHG emissions......................................65! 3.3.2.! Classifying auto-rickshaws as ‘high-PM emitters’ ..................................................69! 3.4.! Conclusions .....................................................................................................................71! 3.5.! References .......................................................................................................................73! 4.! Chapter 4: Climate and health relevant emissions from in-use Indian three-wheelers fueled by natural gas and gasoline.............................................................................................75! 4.1.! Introduction .....................................................................................................................75! 4.2.! Experimental methods.....................................................................................................78! 4.2.1.! Vehicles and fuels.....................................................................................................78! 4.2.2.! Data collection and analysis ....................................................................................79! 4.3.! Results and discussion.....................................................................................................80! 4.3.1.! Emission rates and fuel consumption.......................................................................80! 4.3.2.! Real-world emissions vs. norms ...............................................................................84! 4.3.3.! Climate impacts of CNG...........................................................................................85! 4.3.4.! CNG engines: 4-stroke good, 2-stroke bad? ............................................................86! 4.3.5.! Assessing CNG for auto-rickshaws in Delhi ............................................................87! 4.4.! References .......................................................................................................................89! 5.! Chapter 5: Fuels, technology and vehicle maintenance: Assessing strategies to reduce emissions from Indian auto-rickshaws ......................................................................................92! 5.1.! Introduction .....................................................................................................................92! 5.2.! Methods: Quantifying health and climate-relevant emissions ........................................94! 5.2.1.! Vehicles and fuels.....................................................................................................94! 5.2.2.! Calculating climate impacts.....................................................................................96! 5.3.! Modeling emission control strategies..............................................................................97! 5.3.1.! Replace 2-stroke engines..........................................................................................98! 5.3.2.! Switch from gasoline to alternative fuel...................................................................99! 5.3.3.! Inspection/maintenance program (Type I): Idle emissions testing ..........................99! 5.3.4.! Inspection/maintenance program (Type II): Chassis dynamometer testing ..........100! 5.3.5.! Scrapping old vehicles............................................................................................101! 5.4.! Results and discussion...................................................................................................101! 5.4.1.! Impact of policies on emission factors ...................................................................101! 5.4.2.! Climate considerations...........................................................................................106! 5.4.3.! Integrating AQ and climate co-benefits .................................................................107! 5.5.! Ranking policy options..................................................................................................108! 5.6.! Conclusions ...................................................................................................................110! 5.7.! References .....................................................................................................................111! 6.! Chapter 6: General Conclusions .......................................................................................114! vi  6.1.! 6.2.! 6.3.! 6.4.! 6.5.! 6.6.!  Summary of thesis objectives........................................................................................114! Fieldwork: The Indian Auto-Rickshaw Project (IARP)................................................115! Synthesis of thesis findings ...........................................................................................118! Limitations of the dissertation.......................................................................................120! Policy implications of the research ...............................................................................122! References .....................................................................................................................125!  Appendices .................................................................................................................................126! A.! Appendix A: Supporting information for Chapter 2 .....................................................127! A.1.! Supporting information: Background...........................................................................127! A.2.! Supporting information: Methods ................................................................................128! A.2.1.! Emission factors.....................................................................................................128! A.2.2.! Global warming/cooling metrics for climate forcing aerosols..............................131! A.3.! Supporting information: Results...................................................................................132! A.4.! Supporting information: References.............................................................................133! B.! Appendix B: Supporting information for Chapter 3......................................................135! B.1.! UBC Behavioural Research Ethics Board: Approval/Renewal certificates .................135! B.2.! Survey consent form for auto-rickshaw drivers (English and Hindi)...........................137! B.3.! Survey instrument for ‘Three-Seater Rickshaw’ (TSR) drivers ...................................139! C.! Appendix C: Supporting information for Chapter 4 .....................................................151! C.1.! Emissions regulations and test vehicles........................................................................151! C.2.! Supporting information: Methods.................................................................................153! C.2.1.! Details of test protocol and emission measurements.............................................153! C.2.2.! Details regarding data analysis.............................................................................154! C.3.! Supporting information: Results...................................................................................158! C.4.! Supporting information: References.............................................................................163! D.! Appendix D: Supporting information for Chapter 5 .....................................................165!  vii  LIST OF TABLES Table 1.1. Global Warming/Cooling Potentials (GWP/GCP) for the climate-forcing constituents of motor vehicle exhaust. ..............................................................................................................15! Table 2.1. Summary of climate-forcing emissions factors for liquid-fuel and CNG public transportation vehicles: gaseous and aerosol species. ...................................................................47! Table 2.2. Net present value of carbon credits due to replacing retrofitted CNG engines with new (or improved) CNG engines. Two mitigation options are shown, corresponding to 45% and 100% reduction in CH4 emissions factors, respectively................................................................56! Table 3.1. Specifications for spark-ignited auto-rickshaws fueled with compressed natural gas (CNG)............................................................................................................................................62! Table 3.2. Variables collected in the survey and included in the logistic regression model. .......65! Table 3.3. Vehicle activity, fuel consumption and CO2 emissions for CNG-fueled autorickshaws in Delhi. ........................................................................................................................68! Table 3.4. Results of logistic regression analyses, according to the different groupings analyzed. .......................................................................................................................................................70! Table 4.1 Characteristics of vehicle groups in chassis dynamometer testing...............................78! Table 5.1. Fuel consumption and distance-based emissions factors for the base datasets; distribution in data is represented by median and inter-quartile range. ........................................95! Table 5.2. Model implementation criteria for each strategy to reduce auto-rickshaw emissions.98! Table 5.3. Overview of AQ-climate policy assessment. Solid boxes give examples of policies that should be pursued, while dashed boxes give examples of less effective policies................108! Table 5.4. Reduction in PM2.5 emissions (tonnes per 5,000 auto-rickshaws per annum) attributable to a given policy. ......................................................................................................109! Table A.1. Public transportation vehicles in Delhi.....................................................................127! Table A.2. Average fuel consumption and CO2 emissions factors.............................................127! Table A.3. ‘Leakage’ methane emissions factors (g/km) related to compression of the natural gas, vehicle refueling, and ‘evaporative’ emissions from the CNG vehicles’ fuel systems. ......129! Table A.4. Global warming/cooling potentials...........................................................................131! Table A.5. Total annual climate-forcing emissions inventory. ..................................................131! Table C.1. Indian mass emission limits for three-wheeled vehicles, as measured on the Indian Drive Cycle. As of April 2010, Bharat Stage IV Norms have been introduced in 11 major cities across India. Bharat Stage IV is equivalent to Euro 4 (implemented in Europe in 2005). .........152! Table C.2. Specifications of spark-ignited auto-rickshaws fueled with compressed natural gas (CNG) or gasoline/petrol (PET), with rrear-mounted engines in either 2-stroke or 4-stroke configurations..............................................................................................................................152! Table C.3. Velocity table for one 108-second sub-cycle of the Indian Drive Cycle..................157!  viii  Table C.4. Fuel-based emission factors for the 41 vehicle tests in this study. Expressed uncertainties in EFs are the result of error analysis based on instrument uncertainties. .............158! Table C.5. 100 year global warming potential (GWP100)...........................................................160! Table C.6. Auto-rickshaw fuel consumption and fuel-based emission factors (mean and 95% CI) of gaseous air pollutants. Vehicles are grouped by engine type (4-stroke or 2-stroke), fuel (CNG or gasoline/petrol), and age (‘new’: 2007-2009, or ‘old’: 1998-2001) .......................................160! Table C.7. Fuel-based emission factors (mean and 95% CI) for fine particulate matter (PM2.5), organic carbon (OC), elemental carbon (EC) and GWC. Vehicles are grouped by engine type (4stroke or 2-stroke), fuel (CNG or gasoline/petrol), and age (‘new’: 2007-2009, or ‘old’: 19982001)............................................................................................................................................161! Table C.8. Comparison of distance-based emission factors from this study against other studies. .....................................................................................................................................................162! Table D.1. Distance-based emission factors for auto-rickshaws with 2-stroke engines: base datasets and policy-impacted groups (uncertainty shown by median and inter-quartile range). 165! Table D.2. Distance-based emission factors for auto-rickshaws with 4-stroke engines: base datasets and policy-impacted groups (uncertainty shown by median and inter-quartile range). 166!  ix  LIST OF FIGURES Figure 1.1. Three-wheeled vehicles in India, manufactured by Bajaj Auto Ltd. A: Passenger auto-rickshaw in Delhi, fueled by natural gas. B: Three-wheeled goods carrier in Pune...............3! Figure 1.2. The on-road vehicle fleet in India. A: Growth in the total number of vehicles and by vehicle category, 1951-2004; growth rates are given in parentheses in the legend. B: Proportions of vehicles in the main categories (2004): motorized two-wheelers, light-duty four-wheelers (cars, light trucks and vans), heavy-duty diesel vehicles (buses and tricks), and ‘other’, which includes three-wheeled auto-rickshaws and tractors (MORTH 2006). ...........................................9! Figure 1.3. Schematic of the factors that influence the mass (and composition) of tailpipe emissions from motor vehicles......................................................................................................18! Figure 1.4. Particulate matter (PM) mass emissions standards for new heavy-duty vehicles in different regions over time. Emissions standards in developing countries lag those in the developed world. Note that some cities in India/China (e.g., Beijing, Delhi and other major Indian cities) must meet the standards earlier than the nationwide requirements.........................20! Figure 1.5. Conceptual diagram of co-impacts framework for emissions control policies – health (PM-reduction) or climate (GWC) policies can all be located in this space. A: Quadrant 1 (Q1) is the only space that has both climate and health co-benefits. The shift to point ‘X’ (dashed arrow) in Q2 represents an emissions-related policy that has a health benefit but a climate dis-benefit. ‘Health benefit’ means a cut in PM emissions (leading to reduced ambient PM concentrations and less human exposure); ‘climate benefit’ means a reduction of CO2 equivalent emissions (both GHGs and PM). B: The hypothetical co-impacts of various emission-reduction options. The size of the ovals represents the authors’ subjective uncertainty space for each policy/technology option. Considering other aspects of decisions (e.g., cost and feasibility) would add further ‘dimensions’ to this framework (not shown on this figure). ...........................29! Figure 2.1. Emission inventories, demonstrate the change in climate-forcing emissions attributable to the switch from diesel- and gasoline-fueled vehicles to CNG vehicles. Units are 103 tons of CO2 equivalent emissions (CO2-eq). A. All climate forcing emissions, including black carbon, particulate organic carbon and sulfur dioxide (precursor to sulfate particulate) aerosol species, are included. B-D. Change in climate-forcing emissions (!CO2-eq) due to fuelswitching buses, cars and auto-rickshaws respectively. Note that the vertical scale on panel B is twice that of panels C and D. ........................................................................................................51! Figure 2.2. Sensitivity of model results to emission factors: effect on change in net CO2-eq emissions when CH4 and PM emissions factors are varied around the ‘medium’ emissions factors used for the detailed analysis described in section 2.3.1 (indicated here by the black point in the centre of the figure). The x and y axes refer to CH4 and PM emissions factors respectively. The contours on the graph indicate the percentage change in CO2-eq after the CNG switch, for the range of CH4 and PM emissions factors tested (‘low’, ‘medium’ and ‘high’ values), and the shading shows the area of net climate benefit (CO2-eq reduction). ..............................................54! Figure 3.1. Vehicle model year frequency distribution by engine type (N = 349).......................66! Figure 4.1 Fuel-based emission factors for gaseous pollutants, for 4-stroke CNG and gasoline/petrol-fueled auto-rickshaws (CNG-4S and PET-4S) and for 2-stroke CNG-fueled autorickshaws (CNG-2S): A: carbon monoxide, B: total hydrocarbons, C: methane, D: oxides of x  nitrogen, and E: carbon dioxide, with lines indicating upper limits on CO2 (for ideal combustion). Tukey box plots are used: the line in the box indicates group median, box ends give the interquartile range (IQR) and whiskers show data within 1.5 * IQR of the box end; outliers beyond this range are indicated by data points. NOX was the only species with EFs found to differ significantly (95% confidence interval) between ‘new’ (MY 2007-2009) and ‘old’ (MY 2000-2001) vehicle age groups. PET-4S vehicles have therefore been disaggregated by age in panel D; data for the PET-4S ‘old’ group are shown as individual points because of the small sample size (N=4)..........................................................................................................................81! Figure 4.2 Fuel-based particulate matter emission factors (A: PM2.5 mass, B: organic carbon fraction [OC/PM], C: elemental carbon fraction [EC/PM]), global warming commitment (D: GWC-Kyoto, E: GWC-All), and F: fuel consumption. PM2.5 emissions from CNG-2S vehicles are an order of magnitude larger than from the 4-stroke groups, so the data have been scaled by a factor of ten (Panel A). See Fig. 4.1 caption for box plot description. .........................................82! Figure 5.1. Base emission factor data and modeled policy-induced change in distance-based pollutant emissions factors. A. PET-2S vehicle group. B. CNG-2S vehicle group. C. PET-4S vehicle group. D. CNG-4S vehicle group. Note that the x-axes are not all at the same scale, so – to facilitate comparison across the panels – the number ‘10’ has been highlighted in red on each x-axis. Also, PM2.5 and NOX emission factors have been multiplied by a factor of 100 and 10, respectively, to make them possible to read on the same scale...................................................103! Figure 5.2. Group average global warming commitment (g CO2-equivalent per km) for each fuel/engine type and policy. All climate-forcing agents are shown, including short-lived species (CO, NMHC, EC and OC) as well as the Kyoto protocol species (CO2 and CH4).....................107! Figure 6.1. Indian Auto-rickshaws being tested at the International Center for Automotive Technology in Manesar, near New Delhi....................................................................................118! Figure A.1. Sensitivity of model results.....................................................................................132! Figure C.1. Sub-cycle from the Indian Drive Cycle (IDC); this sub-cycle lasts 108 seconds, including idling, and is repeated 6 times to make up the complete IDC.....................................156! Figure D.1. Assessment of emission-control policies using the health-climate framework. Errorbars include uncertainty in emission factors (both axes) as well as uncertainty in global warming potential for climate-forcing species (y-axis). A. All policies. B. Policies applied to vehicles with 4-stroke engines only, with the x-axis covering a smaller range so the impact of different policies can be distinguished. Note that the y-axes in both panels are to the same scale to facilitate comparison. .................................................................................................................................167! Figure D.2. Annual reduction in PM2.5 emissions from a fleet of 5000 auto-rickshaws (tonnes per year). A. Auto-rickshaws with 2S engines. B. Auto-rickshaws with 4S engines. Each axis within a radar plots has the same scale, to show the relative effectiveness of policies. However the axes on Panels A and B are on different scales, so 10 tonnes per year has been highlighted in red font to facilitate comparisons. ...............................................................................................168!  xi  LIST OF ABBREVIATIONS  2-stroke  Type of engine that completes the combustion cycle in two movements of the piston: 1. exhaust/intake; 2. compression/power; also referred to as ‘2S’ (compare 4-stroke). Spark-ignited 2-stroke engines are commonly used to power twowheelers (motorcycles and mopeds) and three-wheelers (auto-rickshaws).  4-stroke  Type of engine that completes the combustion cycle in four movements of the piston: 1. exhaust, 2. intake, 3. compression, 4. power; also referred to as ‘4S’ (compare 2-stroke). Spark-ignited 4-stroke engines are the most common engine type in cars, and are also frequently used in two-wheelers.  ABC  Atmospheric Brown Clouds: predominantly carbonaceous aerosols suspended in the upper atmosphere, have both regional and global climate impacts  AQ  Air Quality  BC  Black Carbon: component of PM defined optically, typically by measuring the light absorbance by particles (compare EC)  CH4  Methane  CFA  Climate Forcing Agent; this term is comparable to greenhouse gas except it includes atmospheric aerosols such as particles containing sulphate, nitrate, elemental carbon, etc. (compare GHG)  CI  Compression Ignition: engine type fueled with diesel or bio-diesel; also referred to as a Diesel engine  CNG  Compressed Natural Gas (mostly comprised of methane)  CO  Carbon Monoxide  CO2  Carbon Dioxide  CO2-eq  Carbon Dioxide-equivalent emissions, calculated by multiplying the mass of a climate forcing agent (CFA) by its global warming potential (GWP), and used in the ‘Global Warming Commitment’ (GWC) Framework to compare the climate impact of different constituents of combustion emissions  DOC  Diesel Oxidation Catalyst  DPF  Diesel Particulate Filter  EC  Elemental Carbon: the inorganic carbonaceous component of particulate matter, as measured by observing the refractory behavior of PM (compare BC) xii  EF  Emission Factor; the mass of an emission (exhaust constituent) emitted per vehicle kilometers traveled or per mass of fuel burned  GDP  Gross Domestic Product  GHG  Greenhouse Gas (compare CFA)  GWC  Global Warming Commitment: the combined warming effect of climate-forcing agents co-emitted during a specific activity  GCP  Global Cooling Potential: an estimate of the total negative radiative forcing relative to carbon dioxide for a mass of climate forcing agent released, over a specified time-period (compare GWP)  GWP  Global Warming Potential: an estimate of the total positive radiative forcing relative to carbon dioxide for a mass of climate forcing agent released, over a specified time-period. By convention, the time horizon over which GWP is calculated is 100 years, though 20-year and 500-year GWPs are commonly presented in the literature (compare GCP)  HDV  Heavy-Duty Vehicle (such as buses an trucks); usually powered by a dieselfueled, compression-ignition engines  I/M  Inspection and Maintenance: a program to identify high-emitters in a fleet of vehicles through regular, mandatory emission testing (‘inspection’), which requires those vehicles that exceed the limits to be repaired or scrapped (‘maintenance’)  IARP  The Indian Auto-Rickshaw Project (the overarching title for the project that included the studies described in this dissertation)  ICAT  International Center for Automotive Technology; a vehicle testing and emission homologation centre in Manesar, India  IITD  Institute of Technology, Delhi  IPCC  Intergovernmental Panel on Climate Change  LDV  Light-Duty Vehicle (includes four-wheeled passenger vehicles such as cars and light trucks, as well as three-wheelers, motorcycles and scooters); usually powered by gasoline-fueled, spark-ignition engines, which may be of 4-stroke or 2-stroke design  LLCFA  Long-Lived Climate Forcing Agent (such as CO2)  LPG  Liquefied Petroleum Gas (mostly comprised of propane) xiii  NMHC  Non-Methane Hydrocarbons  NOX  Oxides of Nitrogen (NO plus NO2)  N2 O  Nitrous Oxide, a greenhouse gas produced in relatively small quantities from vehicle combustion  OC  Organic Carbon  OECD  Organisation for Economic Co-operation and Development  OM  Organic Matter component of PM; for vehicle emissions, organic matter is related to organic carbon by a factor of 1.2 (OM = 1.2 * OC)  PIC  Product of Incomplete Combustion  PM  Particulate Matter. PM is differentiated by aerodynamic diameter (see PM2.5 and PM10)  PM2.5  PM with an aerodynamic diameter of less than 2.5 microns; also known as ‘Fine PM’ or ‘Respirable PM’  PM2.5-10  PM with an aerodynamic diameter of between 2.5 and 10 microns; also known as ‘Coarse Fraction PM’  PM10  PM with an aerodynamic diameter of less than 10 microns; also known as ‘Inhalable PM’  PUC  ‘Pollution Under Control’, India’s inspection and maintenance program for in-use vehicles.  RPT  Regenerative Particulate Trap  SI  Spark Ignition: engine type usually fueled with gasoline; other common fuels used in SI engines include ethanol, compressed natural gas and liquefied petroleum gas  SLCFA  Short-Lived Climate Forcing Agent (such as PM)  THC  Total Hydrocarbon emissions  TRIPP  Transport Research and Injury Prevention Programme at IITD  TWC  Three-Way Catalyst  ULSD  Ultra-Low Sulphur Diesel  UNFCCC  United Nations Framework Convention for Climate Change  xiv  ACKNOWLEDGEMENTS  First of all, I would like to thank my advisor Milind Kandlikar for his mentorship and encouragement throughout my PhD. His enthusiasm about the interesting and valuable work that can be done on social and environmental problems, and his confidence that quantitative interdisciplinary work can contribute to lasting solutions, is contagious and inspiring. I’d like to acknowledge my advisory committee for the insights and constructive critiques they have brought to my program of research: Hadi Dowlatabadi, Steve Rogak and Madhav Badami each offered a different and fresh perspective that helped me improve this dissertation. Many thanks to Andy Grieshop for his essential part in making the fieldwork in India such a success, and for being such a good friend since his arrival in Vancouver. This research was funded by: the AUTO21 Network of Centres of Excellence; the BC Environmental and Occupational Health Research Network; a post-graduate fellowship from the National Sciences and Engineering Research Council; and a strategic training fellowship from the UBC Bridge Program. I am grateful to these institutions for their support. I would also like to acknowledge generous awards from Engineers Canada (the TD Meloche Monnex Scholarship), the Robert B. Caton Memorial Scholarship, the Air & Waste Management Association Graduate Scholarship, the Transportation Association of Canada Foundation Scholarship, and the Les Lavkulich Scholarship. My fieldwork in India would not have been possible without the support of many people and institutions. Many thanks to Suparna Deb Chaudhuri, Manna and Zorro for hosting me at their home in Delhi. Geetam Tiwari and Dinesh Mohan graciously gave me desk space in the Transportation Research and Injury Prevention Programme at the Indian Institute of Technology, Delhi. Rajendra Ravi and his staff at the Institute for Democracy and Sustainability in Delhi provided valuable feedback during development of the driver survey, and were instrumental in conducting the survey and later recruiting drivers and vehicles for the laboratory testing of autorickshaw emissions. V. Sankaran and his team at the International Centre for Automotive Testing (ICAT, Manesar) generously provided in-kind support and facilitation of the emissions measurement project. Thanks also to Julian Marshall for lending me the DustTrak instruments. I am grateful to N.V. Iyer for making the time to answer many questions about three-wheelers in India. Thanks to Adrian Carney for her patience and perseverance with sorting out the financial side of the research in India. The UBC School of Environmental Health and the Center for Atmospheric Particle Studies (CAPS) at Carnegie Mellon University generously provided laboratory and equipment access that were essential for data analysis. I am very grateful to all of the faculty, staff and students in the Institute for Resources, Environment and Sustainability and the UBC Bridge Program for offering such a welcoming and engaging environment to pursue my studies. The many conversations that I have had with colleagues by the espresso machine have been an essential part of my (over-caffeinated) doctoral experience. My experience as a PhD student was enriched by my great friends both at the university and beyond. At UBC I would especially like to thank Christian Beaudrie, Tom Berkhout, Zosia Brown, Matt Brown, Christina Cook, Negar Elmieh, Julian Freeman, Brian Gouge, Tom Green, Anne Harris, Sonja Klinsky, Jana Kotaska, Reza Kowsari, Nico Lhomme, Eric Mazzi, Francis Ries, Charlie Wilson, and Meghan Winters. Thanks to all my friends in xv  Vancouver who share my love of spending time in the outdoors – exploring the wild places of British Columbia with you helped to remind me how important it is to have balance in life. My fellowship in the UBC Bridge program has contributed to my academic balance: I would like to thank Kay Teschke for offering me the opportunity to help develop a research proposal about cycling injuries as part of the Bridge fellowship program, and then for inviting me to continue as a member of the research team. It was great to be able to work on this topic because cycling is very important to me, and the project was a perfect complement to my thesis research. Christie Hurrell has been a huge help with communicating my research findings to a wider audience, for both my thesis research and the cycling work. My parents have been fantastic role models and I really appreciate that they have always supported the choices I have made in life and career: thank you for your love and encouragement, and for trail-blazing our family’s connection with Vancouver. Lastly, my deepest love and gratitude goes to my wife Rebecca Goulding, who has been my constant companion, soul-mate and greatest supporter throughout this endeavour. Thanks for helping me keep things in perspective, for your wonderful sense of humour, and for your love of life – I’m looking forward to sharing many more adventures with you.  xvi  DEDICATION  To my grandfather, Murrogh Vere OBrien  xvii  1. Chapter 1: General Introduction  1.1. Context: Air pollution and emissions from vehicles in India  1.1.1. Motivation The research described in this dissertation is motivated by the extremely poor air quality in the cities of many developing countries in Asia. Emissions from on-road transportation sources are an important contributor to air pollution and climate change. This dissertation is an assessment of certain policies that have been implemented to control vehicle emissions in India. Specifically, the research has examined the climate and air-quality impacts of large-scale adoption of compressed natural gas (CNG) as an alternative to diesel and gasoline. CNG fuelswitching policy has been evaluated in the context of different engine technologies and vehicles of varying age, compared against institutional inspection and maintenance programs. The focus has been on how emissions are affected in real-world vehicles, rather than new vehicles. Data has been collected from vehicles and drivers in Delhi, where compressed natural gas is mandated for all public transport vehicles. However the research design is such that the findings are applicable to many other jurisdictions in the developing world where CNG is being considered as an alternative fuel. Annual mean levels of inhalable particulate matter (PM10) air pollution in three of the largest Indian cities (New Delhi, Kolkata, and Mumbai) are regularly more than double the limit recommended by the World Health Organization, and an order of magnitude higher than most cities in developed countries (Anderson et al. 2004; Molina and Molina 2004). Although there are some cities in North America and Europe where air quality issues remain a significant concern, (e.g., Los Angeles), considerable effort has been expended to understand and control air pollution. In Asian ‘megacities’ such as Delhi, a range of air quality policies have also been implemented, but emission rates from many sources – and air pollution concentrations – remain high (Gurjar et al. 2004; Kandlikar 2007).  1  The rate of growth in transportation activity and energy use per capita in the developing world is driven by the increasing demand for personal mobility and goods movement, which in turn is associated with growing incomes (Dargay et al. 2007). In India, traffic-related problems in urban areas – including pollutant emissions, congestion, noise, and fatalities/injuries – have been described in aggregate as an “urban transport crisis” (Pucher et al. 2005). Indeed, the transportation sector is a critical and rapidly increasing contributor to air pollution in India, at regional as well as local scales (Guttikunda et al. 2005; Gurjar et al. 2004). Yet important knowledge gaps remain related to climate and health-relevant emissions. The research conducted for this thesis aims to address these data gaps by examining policies to reduce tailpipe emissions from transportation in Delhi.  1.1.2. Why study three-wheeled auto-rickshaws in Delhi? For the experimental portion of this dissertation, I have quantified emissions and activity from Indian three-wheeled auto-rickshaws. Three-wheeled vehicles are used in many South Asian countries (particularly India, Pakistan and Bangladesh) for transporting goods as well as people (Figure 1.1). In this thesis I have specifically examined passenger-carrying auto-rickshaws that were operating in Delhi and the surrounding area. Indian auto-rickshaws have a simple, lightweight chassis with open sides, a canvas top and motorcycle-style engine and controls. A bench seat behind the driver has room for 3 passengers. Auto-rickshaws operate as low-cost taxis and are available throughout urban and peri-urban areas, but despite their ubiquity in Indian cities, few previous studies have assessed of this mode. They fill an important niche in developing cities between private vehicle ownership and fixed-route public transit systems (i.e., bus and metro). Auto-rickshaws are likely to remain important even with improvements in public transportation and rising private vehicle ownership, because they provide inexpensive mobility over short and long distances at any time of the day.  2  !"  #"  Figure 1.1. Three-wheeled vehicles in India, manufactured by Bajaj Auto Ltd. A: Passenger auto-rickshaw in Delhi, fueled by natural gas. B: Three-wheeled goods carrier in Pune.  Current estimates of the auto-rickshaw population in India stand at between 3 and 3.5 million, and the worldwide total is estimated to be approximately 4.5 million (Iyer 2003). Autorickshaws in Delhi are estimated to cover around 150 km per day (based on data collected for this thesis; see Chapter 3), which translates to about 55,000 km annually. Since these high-mileage vehicles have simple, single-cylinder engines and a reputation for being poorly maintained, it is reasonable to believe that they emit significant amounts of pollutants and climate-relevant emissions. Although it is difficult to obtain data on the distribution of auto-rickshaws by engine and fuel type, informal observational evidence gathered in many Indian cities suggests that 2-stroke engines are far more prevalent than 4-stroke engines (Kandlikar, M.; personal communication). Carbureted 2-stroke engines have a reputation of being highly polluting, due to their operation on a mixture of fuel and lubricating oil. Autorickshaws are primarily operated within densely populated areas, and physically close to pedestrians and passengers of other vehicles that are open to the environment (i.e. all twowheelers, three-wheelers, and almost any vehicle without air conditioning, since they have their windows open for ventilation). Thus the fraction of their exhaust that might be inhaled (the ‘intake fraction’) is likely to be high relative to other sources, such as power plants or even heavy-duty goods vehicles that operate on inter-city routes, because of their proximity to the  3  exposed population (Bennet et al. 2002). Self-pollution of the driver and passengers within any given auto-rickshaw may also be a problem, especially while idling in traffic. Another reason why these vehicles were excellent candidates for research is that they are available with a range of fuel systems and engine options in an otherwise identical chassis. This enabled a comparison of emission factors across different fuels and engine types, and the determination of variation within our sample of in-use vehicles. The engines in auto-rickshaws are similar to those used in many other kinds of light-duty vehicles in the developing world, including two-wheelers and small cars. This fact increases the usefulness of measuring fuelbased emission factors from auto-rickshaws, because they can be expected to be similar to other vehicle types. Finally, being ‘for-hire’ vehicles, it was relatively straightforward to recruit a sample of vehicles that could be assumed to be representative of the fleet in Delhi.  1.1.3. Links between air quality and climate change Air pollution and climate change are inextricably linked, because both are driven by emissions from combustion. Their impacts differ significantly both spatially and temporally, however. At its most basic level, poor air quality in cities results in direct adverse health impacts for the exposed population, and exposure/health effects can be acute (time scale of minutes to days) and/or chronic (time scale of months to years) (HEI 2010). Climate change, through changing temperature and precipitation, may affect populations far from the source of emissions, and over time scales of tens to hundreds of years (IPCC 2007). There are important similarities, too: those most likely to be affected tend to be the most vulnerable people in less-developed countries, who are least able to modify their behaviour or change location to mitigate exposure to risk (Campbell-Lendrum and Corvalán 2007; O'Neill et al. 2008). The policies to improve air quality and mitigate climate change have traditionally been separately conceived and implemented, because each has focused on different constituents of combustion emissions: long-lived species for climate change (such as carbon dioxide, CO2, or methane, CH4) and short-lived species for air quality (such as atmospheric aerosols, or particulate matter, PM). Increasingly, however, it is recognized that short-lived particulate species contribute significantly to climate forcing as well as health impacts (Ramanathan and Carmichael 2008, Grieshop et al. 2009, Smith et al. 2009), and some relatively long-lived 4  species such as CH4 can affect regional air quality (West et al. 2006). It is this overlap in climate and health impacts that leads to possibilities for co-benefits. It has been claimed that climate mitigation policies will universally result in air quality benefits, because efforts to reduce CO2 will be driven by ‘clean combustion’ technologies, efficiency improvements, and developments of alternatives to fossil fuel and biomass use (Bell et al. 2008; Smith and Haigler 2008). Unfortunately it has been demonstrated that otherwise-effective climate policies may cause additional air pollution and an increase in adverse health effects if they are not appropriately coordinated with air quality policies (Mazzi and Dowlatabadi 2007). For climate policies to succeed, it can be argued that it is necessary to simultaneously pursue airquality co-impacts (and vice-versa). There are many opportunities for co-benefits, especially in developing countries, but to date insufficient attention has been paid to quantifying transportation-related opportunities.  1.1.4. Research scope and strategy For this dissertation, I have quantitatively examined the impact of certain strategies to reduce tailpipe emissions from auto-rickshaws and other vehicles in Delhi, specifically: (1) The Indian Supreme Court-mandated requirement for all public transport vehicles in Delhi to switch from diesel and gasoline to CNG. (2) Mandatory ‘inspection and maintenance’ (I/M) programs, which check that light-duty inuse vehicles are meeting emissions standards during an engine-idle test. (3) Phasing out of vehicles using 2-stroke engine technology, and the consequent increase in the use of 4-stroke engines (and the impact of CNG use in both engine types). This research is policy-driven: it quantitatively assesses the impacts of policies and provides recommendations for policy-makers. The work is also highly interdisciplinary, drawing on a broad range of literatures in an attempt to gain a clearer understanding of the scope of the problem, including: environmental science (measurement of vehicle emissions, ambient air pollution sources and concentrations, and climate forcing of different combustion-related emissions); public health (air pollution epidemiology); and, integrated assessment (methodologies to gain insights into complex environmental problems that span a wide range of spatial and temporal scales, and cross disciplinary boundaries). 5  The overall objective of this dissertation was to quantify the effectiveness of emission control policies for in-use vehicles in India, motivated by the urgent need for reliable data about transportation emissions in developing countries. It is very difficult to collect empirical data about vehicle activity and emissions, because there is huge variation in the types and condition of on-road vehicles, and researchers and policy analysts usually have limited resources. This doctoral thesis is composed of a series of related empirical research projects, based in India. I have studied the impacts of the large-scale adoption of CNG as an alternative to diesel and gasoline in New Delhi, and more specifically I have conducted focused empirical studies of the activity patterns and emissions of three-wheeled auto-rickshaws. The thesis combines multiple methods to meet its objectives, including: a structured survey of auto-rickshaw drivers and associated social science methods; laboratory-based empirical data collection; policy analysis and integrated assessment. The overall empirical research project has been given the title The Indian Auto-Rickshaw Project (IARP), because the research has primarily involved examining the three-wheeled autorickshaws that are ubiquitous in Asian cities. Fieldwork was conducted over three trips to New Delhi, in 2007, 2008 and 2009, each of around two months duration. During the first trip, I conducted one-to-one meetings with over 20 experts from academia, government, NGOs and industry, all of whom had expertise in the field of urban transportation in India. Critical to the success of the subsequent trips and research projects in IARP, I worked with an NGO called the Institute for Democracy and Sustainability (ISD), which was instrumental in facilitating the survey in 2008, and also for recruiting a sample of in-use auto-rickshaws for laboratory emissions testing in 2009. I conducted the survey during my second trip to Delhi, and collected information from 350 drivers. Testing and Measurement of auto-rickshaws was done at the International Center for Automotive Testing (ICAT) in Manesar, near New Delhi in Fall 2009. Analysis of samples was done at the Center for Atmospheric Particle Studies (CAPS) at Carnegie Mellon University, in Pittsburgh. Further discussion of the process by which this program of research, IARP, was designed, developed and implemented can be found in the general conclusions of this thesis (Chapter 6). There, I also discuss a number of interesting ‘spinoff’ projects which are part of IARP, but not part of this thesis.  6  Where critical knowledge gaps were identified as part of the IARP project development, primary data was collected using auto-rickshaws as the test-subject. Assumptions have been clearly documented, and uncertainty has been dealt with explicitly using sensitivity analyses and propagation of uncertainty wherever possible. While this chapter provides a qualitative background to the challenge of achieving health and climate co-benefits from ‘emission reduction’ policies, subsequent chapters in this thesis describe the findings of quantitative analyses. Overall, the aim of this work was to integrate a range of quantitative methods to provide insights for policy-makers in the transportation, air pollution and climate domains. In the remainder of this chapter, I will identify the specific research objectives for each chapter (1.2). I then provide a review of the emissions that are produced from fossil fuel combustion in motorized vehicles and summarize their air quality/health and climate change impacts (section 1.3). In this section I will also discuss ‘emission inventories’, because they are at the core of quantitative assessments of transportation impacts. The next section (1.4) gives an overview of the principal policy approaches that can be used to control emissions from vehicles (with a focus on India), and then a simple framework is presented (section 1.5) for assessing these approaches according to their potential to reduce climate and health impacts. Finally, I give an overview of the dissertation structure and outline the principal findings of each chapter (1.6).  1.2. Research objectives  The overall objective of this research was to quantify the impacts of emission control policies (including switching fuels and engine technologies) on emissions from in-use vehicles in India. In particular, I focussed on understanding the impacts of the large-scale adoption of compressed natural gas (CNG) as an alternative to diesel and gasoline in Delhi. Specific sub-objectives of this research, which are addressed in individual chapters, were to: •  Quantify the climate impacts of a fuel-switching policy (CNG in Delhi), with particular attention to the radiative forcing effects of short-lived climate forcing agents, i.e., atmospheric aerosols (Chapter 2).  •  Determine activity factors, fuel consumption and CO2 emissions for natural gas-fueled auto-rickshaws through a structured survey of drivers, and develop a model to examine  7  potential determinants of ‘high-PM emitters’, i.e., those vehicles emitting visible smoke (Chapter 3). •  Establish gaseous and fine particulate emission factors for spark-ignited auto-rickshaw engines fueled with natural gas and gasoline, by conducting chassis dynamometer testing on in-use Indian auto-rickshaws (Chapter 4).  •  Synthesize the findings from the previous studies in a quantitative policy analysis, and compare the change in emissions that could be attributed to a range of transportation technologies and policies, and investigate the utility of the AQ-climate ‘co-impacts’ framework (Chapter 5).  This is a ‘manuscript-based’ dissertation, with each research chapter written as a stand-alone manuscript for publication in the peer-reviewed academic literature. Further details of this format are given in Section 1.6, followed by an overview of the research contributions that arose from each of the dissertation’s chapters.  1.3. Transportation energy use and its impacts1  1.3.1. Transportation activity and fuel-use in India Globally, the on-road transportation sector accounts uses more than 60 EJ of petroleum-based fuels each year, which is about 20% of total world energy use (IEA 2006). In 2004, road transport produced 4.7 Gt carbon dioxide (CO2), or about 17% of world energy-related CO2 emissions (Kahn Ribeiro et al. 2007). Singh and colleagues (2008) have estimated that the Indian road transport sector was responsible for 106 Mt of CO2 emissions in 2000, which is about 2% of the world transportation total. Although this number appears relatively small, freight and passenger transportation demands in rapidly industrializing countries such as China and India are growing fast, and merit research attention. India’s annual petroleum consumption in the roadtransport sector (diesel and gasoline) has increased almost fourfold since 1980, to just under 20 million tonnes in 2004 (Singh et al. 2008; TERI 2006). 1  A version of part of this chapter (Sections 1.3-1.5) has been published: Reynolds, C., Grieshop, A. and Kandlikar, M. (in press) “Reducing Particulate Matter Emissions from Buses and Trucks in Asia: A Framework to Assess Air Pollution and Climate Change Co-Impacts”, in Zusman, E., Srinivasan, A. and Dhakal, S. (eds) Low Carbon Transport in Asia: Strategies for Optimizing Co-benefits, Earthscan, London.  8  Vehicle ownership rates are high in Europe and North America, on the order of 400-800 vehicles per 1000 persons, but growth rates are relatively low because the market for private vehicle ownership is close to saturation (World Bank 2010). In contrast, in India the ownership rate is far lower – an estimated 12 vehicles per 1000 persons in 2003 – but there is a very high rate of new vehicle ownership that is driven by rapidly increasing GDP per capita (World Bank 2010; Dargay et al. 2007). Historical growth in total number of motor vehicles in India has been exponential over recent decades, with a doubling period of 6.7 years (MORTH 2007). Figure 1.2.A shows the trend in vehicle registration numbers from the early 1950s to 2004 by mode. Over most of this period, this growth has been driven largely by private purchases of new twowheelers (motorcycles and scooters), which currently make up just over 70% of all Indian vehicles (Fig. 1.2.B). However, 2007-2008 data from the Indian automobile industry suggest that two-wheeler sales are dropping while the growth rate for four-wheeled passenger vehicles is now over 12% (SIAM 2010).  100,000  Mode (annual growth rate)  A  All vehicles (10.4%) Light-duty four-wheelers (8.4%)  10,000  Others, incl. three-wheelers (9.1%) Goods vehicles, trucks (7.4%)  1,000  Buses (6.2%)  100  10  1 1950  1960  1970  1980  Year  1990  2000  No. of vehicles (millions)  No. of Vehicles (thousands)  Two-wheelers (11.3%)  B  60 50 40 30 20 10 0  Year: 2004  Twowheelers  Light-duty Buses and fourtrucks wheelers  "Other"  Figure 1.2. The on-road vehicle fleet in India. A: Growth in the total number of vehicles and by vehicle category, 1951-2004; growth rates are given in parentheses in the legend. B: Proportions of vehicles in the main categories (2004): motorized two-wheelers, light-duty four-wheelers (cars, light trucks and vans), heavy-duty diesel vehicles (buses and tricks), and ‘other’, which includes three-wheeled auto-rickshaws and tractors (MORTH 2006).  9  Only about 6% of the current Indian fleet are heavy-duty vehicles (HDVs; buses and trucks with compression-ignition [CI] engines operating on diesel fuel), but they travel further and have higher fuel consumption than light-duty vehicles (LDVs; cars, light trucks and two- and threewheelers with spark-ignition [SI] engines, mostly fueled with gasoline). Consequently diesel fuel use for transportation in India (~35 Mt oil-equivalent [Mtoe], or ~1.5 EJ) is approximately three times higher than gasoline use (~10 Mtoe, or ~ 0.4 EJ) (Baidya and Borken-Kleefeld 2009). Gaseous alternative fuels – compressed natural gas (CNG) and liquefied petroleum gas (LPG) – are also important transportation fuels in some Indian cities, but represent only about 1% of total fuel consumption on an energy basis (Baidya and Borken-Kleefeld 2009). At the national level, HDVs use more fuel and may produce more pollutant emissions than LDVs, but LDVs are an important source of emissions in urban areas, where most of the population resides. Therefore if both air quality (health) and climate benefits are to be realized, interventions to reduce transportation emissions must consider both heavy- and light-duty modes.  1.3.2. Pollutant emissions from vehicle engines Ideal combustion of hydrocarbon-based fuels would result in only CO2 and water vapour emissions. In real combustion processes, pollutant species are also formed and emitted. Oxides of nitrogen (NOX, i.e., NO and NO2) are formed at high temperatures from the nitrogen and oxygen present in air. Other pollutants are due to incomplete mixing of fuel and air during the intake process, the short combustion time-scale (order of milliseconds), and flame-quenching at the combustion chamber walls (among other things). As a result, products of incomplete combustion (PIC) are formed, which include carbon monoxide (CO), methane (CH4), and nonmethane hydrocarbons (NMHC). Vehicle emissions of NOX, NMHC and CO contribute to the formation of tropospheric ozone (O3), a powerful GHG that also causes respiratory health effects. Other gaseous pollutants that are emitted from engines are air toxics (such as benzene, toluene and 1,3 butadiene) and sulphur dioxide (SO2). In addition to gaseous pollutants, particulate matter (PM) is emitted from both compressionignition and spark-ignition engines. PM is mostly composed of carbon, and is normally formed in fuel-rich regions of the combustion chamber (Kittelson 1998). Other factors that contribute to PM formation include lubricating oil (type and quality), engine wear processes, and fuel impurities (such as sulphur). Motor vehicle PM typically measures on the order of tens to 10  hundreds of nanometres in diameter (Kittelson 1998). Heavy-duty diesel engines are about 20% more fuel-efficient than spark-ignition engines, but they emit far more PM per vehicle km or per mass of fuel burned. For example, a study of vehicular PM in a Californian roadway tunnel found that average PM emissions per mass-fuel-burned were 20-30 times higher for diesel HDVs than for gasoline LDVs (Kirchstetter et al. 1999). On the other hand, in cities there tend to be many more LDVs than HDVs, so their contribution to ambient PM concentrations may be of the same order of magnitude. PM emission rates are very sensitive to transient engine operation (i.e., changes in engine speed and/or load): it has been estimated that the average PM emission rate for diesel engines during transient operation can be between two (Samulski and Jackson 1998) and seven times (Cocker et al. 2004) greater than that during steady-state operation. Other critical factors that can influence vehicles’ PM mass emission rates are vehicle age and condition (factor of 2), weight class (factor of 2), and the drive-cycle used for testing (up to a factor of 10) (Clark et al. 2002; Lough et al. 2007). It has long been realized that a substantial portion of overall emissions can come from a small fraction of vehicles that each emit far more than the average (Beaton et al. 1995; Calvert et al. 1993). Such ‘super-emitters’ typically comprise less than 10% of the fleet, yet they can contribute 50% or more of the emissions (depending on species) due to poor maintenance or failure of some part of the emissions control system (Ban-Weiss et al. 2009; Subramanian et al. 2009). Typically, between 40-80% by mass of primary PM from diesel engines is elemental carbon (EC, also known as black carbon or soot), while the remainder is mostly organic carbon (OC), sulphates, minerals and metals (Gillies and Gertler 2000; Kittelson 1998). The composition of emitted PM plays a central role in determining its climate impacts, and is also likely to influence its toxicity – both aspects will be discussed in greater detail below. Therefore, determination of both the quantity and composition of PM emitted is critical for estimating emissions’ impacts. There is significant inter- and intra-vehicle variability in PM emission rates, especially from diesel vehicles (Yanowitz et al. 2000). For PM from 4-stroke spark-ignition engines, about 2040% of its mass is EC, which is still significant but is less than for diesel engines (Kirchstetter et al. 1999; Chapter 4 of this dissertation). In contrast, PM from 2-stroke spark-ignition engines is dominated by OC from unburned lubricating oil (Sakai et al. 1999; Volckens et al. 2008; Chapter 4 of this dissertation).  11  1.3.3. Air quality and health effects of traffic-related air pollution Worldwide, urban outdoor air pollution is estimated to result in at least 800,000 excess deaths annually (Rodgers et al. 2002). The greater health burden falls on urban populations in rapidlyindustrializing countries such as India, and about 60% of these excess deaths occur in Asia alone (Anderson et al. 2004). PM is thought to be the critical component of ambient air pollution from a public health perspective, though other gaseous exhaust constituents and air toxics also have important health effects (HEI 2010). The ‘fine’ fraction of ambient PM (i.e., PM with aerodynamic diameter of less than 2.5 microns, or PM2.5) is dominated by emissions formed during combustion, and has been linked to cardiovascular, cardiopulmonary, and respiratory mortality and morbidity in numerous North America and European studies (Pope and Dockery 2006). The concentration-response relationship for PM and health impacts has been shown to be remarkably consistent between cities with a wide range of ambient PM concentration; an increasing number of studies confirm that the relationship is similar in Asian cities (e.g., Wong et al. 2008). In this thesis, I focus on PM2.5 reduction when assessing policies, because the adverse health impacts of exposure to air pollution are driven primarily by the PM component, according to current science (Pope and Dockery 2006). PM source apportionment studies in Indian cities estimate that fossil fuel combustion contributes between 20-60% of ambient PM2.5 mass (Chowdhury et al. 2007). Some studies attribute up to half of this to diesel engines (Srivastava and Jain 2008). However there is considerable uncertainty in such source apportionment studies, especially related to parsing the respective contributions of gasoline and diesel fueled vehicles. The largest uncertainty is believed to be associated with the characterization of ‘smoking’ vehicles, or ‘high-PM emitters’ (Lough et al. 2007, Lough and Schauer 2007). As an example of the level of uncertainty common in source apportionment studies, a recent study of on-road vehicles in Mexico City presents estimates (mean, 95% confidence interval) of total PM2.5 from gasoline and diesel vehicles: 210 tonnes yr-1 (0–400) from gasoline vehicles and 620 tonnes yr-1 (260–2,380) from diesel vehicles, not including additional uncertainty related to the PM measurement instrumentation (Thornhill et al. 2010). Relative to higher-income countries, a smaller proportion of ambient PM in developing countries is from vehicles and other internal combustion engines, because other sources such as refuse 12  burning and biomass combustion for cooking and heating make a significant contribution to emissions in both urban and rural areas (Bond et al. 2004). However this should not obscure the fact that fuel-based PM emission rates from the motorized vehicle fleet are likely substantially higher in developing Asia than in higher income countries, due to vehicles being older and more poorly maintained on average (e.g., Subramanian et al. 2009). Since diesel engines have higher PM emission factors (mass emitted per vehicle distance traveled or per distance traveled) than spark-ignition engines, they have received greater attention from researchers and policy-makers to date. Increasingly, however, epidemiologic studies are finding robust connections between ultrafine nanoparticles (PM1.0) and adverse health effects in exposed populations (Delfino et al. 2005). Since spark-ignition engines produce high number concentrations of ultrafines – albeit with relatively little total mass – there is good reason to characterize their emissions and consider regulatory control. To paraphrase one of the leading researchers in the field, there is both bad news and good news related to traffic-related air pollution and health (Pope 2004). The bad news is that pollution from transportation is ubiquitous in urban air and is difficult to control, in large part due to the multitude of sources contributing to ambient concentrations. The good news is that exposure to air pollution is a modifiable risk factor, which means that there are significant opportunities to prevent disease if ways of reducing emissions and exposure can be found. While some countries have been relatively successful in controlling PM levels, most cities in Asia have ambient PM concentrations far exceeding World Health Organization guidelines (Molina and Molina, 2004; Gurjar et al. 2008; Hopke et al. 2008; Kim Oanh et al. 2006). Although attribution to specific sources is challenging, it is clear that a significant proportion of ambient fine PM in Asian cities comes from road transportation sources (Chan and Yao 2008; Chowdhury et al. 2007). Therefore there is much work yet to be done on tackling air pollution from mobile sources to reduce trafficrelated disease risks.  1.3.4. Climate impacts of traffic-related air pollution Exhaust constituents that influence the Earth’s radiative balance are categorized as long-lived or short-lived climate forcing agents (LLCFA and SLCFA, respectively) (Shine et al. 2007). The former category includes gaseous species such as CO2 and CH4, while the latter includes CO, 13  NMHC and atmospheric aerosols (such as sulphates and PM). CO2 is the principal greenhouse gas (GHG) emitted from transportation sources, with other species contributing on the order of 5-10% of radiative forcing (Kahn Ribeiro et al. 2007). The global warming commitment (GWC) of a mass of vehicle exhaust can be calculated using the following equation: n  GWC" = EFCO2 + # EFi $ GWP",i  (Equation 1.1)  i=1  where EFi is the emission factor of exhaust constituent i, and GWP",i is the global warming  !  potential for exhaust constituent i over time-period ! (by convention, ! is usually 100 years). The only exhaust constituents included in the Kyoto Protocol are CO2 and CH4, but the ! Intergovernmental Panel on Climate Change (IPCC) also provides GWP estimates for CO and NMHC (Forster et al. 2007). In addition, atmospheric particles have important – though as yet poorly understood – impacts on the energy balance of the earth and thus on the global and regional climate2. A number of researchers have made estimates of GWP for components of PM (OC and EC) and SO2 (the precursor to sulphates) (e.g., Hansen 2007, Bond and Sun 2005). In this dissertation I use GWP values for EC, as well as ‘global cooling potentials’ (GCP) for OC and SO2, calculated from radiative forcing and atmospheric lifetime information given in the IPCC’s Fourth Assessment Report (Forster et al. 2007; Reynolds and Kandlikar 2008). Table 1.1 gives the GWP100 for the LLCFAs and SLCFAs considered in this research.  2  A detailed discussion of the mechanisms by which atmospheric aerosols cause direct and indirect radiative forcing is beyond the scope of this chapter, but I have co-authored publications that explore this topic in depth (Grieshop et al. 2009, Kandlikar et al. 2009).  14  Table 1.1. Global Warming/Cooling Potentials (GWP/GCP) for the climate-forcing constituents of motor vehicle exhaust. Pollutant  Global Warming/Cooling Potential a  CO2 b  1  CH4  b  25  CO  1.9 (1.0 to 3.0)  NMHC  3.4 (1.7 to 6.8)  PM: EC  455 (190 to 720)  PM: OC  -35 (-10 to -80)  b  SO2 -100 (-50 to -210) 100 year time horizon; uncertainty ranges are given in parentheses, calculated for EC, OC and SO2 from  a  uncertainty in radiative forcing values (Reynolds and Kandlikar, 2008). b  Only carbon dioxide and methane are include in the Kyoto Protocol.  c  Gaseous sulphur dioxide gas is emitted from heavy-duty vehicles burning sulphur-containing fuel, and is the  critical precursor to sulphate aerosols.  Global warming and cooling metrics enable the radiative forcing impacts of SLCFAs to be evaluated against conventional GHGs, but there are undoubtedly challenges to including aerosols in global climate agreements.3 Therefore, in this dissertation I have presented GWC for Kyoto gases only (CO2 and CH4), as well as all climate-forcing exhaust constituents (Kyoto gases plus CO, NMHC, EC, OC and SO2). Despite significant uncertainty in metrics such as GWP/GCP for aerosols (Forster et al. 2007), they are important because they allow vehicular PM to be considered in integrated assessments of emissions-reduction approaches. Sufficient evidence has amassed regarding the adverse climate impacts of atmospheric black carbon to provide another reason (in addition to their adverse health impacts) to reduce vehicular PM emissions (Bond 2007; Swart et al. 2004).  1.3.5. Quantifying emissions from mobile sources: Emission inventories To decide how to reduce emissions from vehicles, it is necessary to first determine the scale of the problem and then to measure progress as policies are implemented. To produce a quantitative emission inventory for a given exhaust constituent from a fleet of vehicles, the level of vehicle  3  Interested readers should consult Bond (2007) for an assessment of some of the barriers to including PM in climate agreements, and arguments applied to overcome them.  15  activity (fuel use or distance traveled, disaggregated by vehicle mode) and the amount of pollution emitted per vehicle activity are used in the following equation: n  E i,t = # Am,t " EFi,m  (Equation 1.2)  m =1  where Ei,t is the total emission from a fleet of exhaust constituent i, over a time period t (usually  !  one year); Am,t is the total activity (typically fuel use, kilometres traveled or kW-hour of engine power output) for vehicle mode m (e.g., cars, two-wheelers, trucks) during time period t; and EFi,m is the emission factor or average mass of pollutant i formed per unit of vehicle activity. Emission factors for pollutant species from engines can vary by orders of magnitude, depending inter alia on: engine design; fuel quality and sulphur content; presence, type and condition of emissions aftertreatment device; age and condition of the engine; engine load; local topography; climatic conditions; and driving patterns. Hence the use of average EF (and average annual vehicle activity) is a significant simplification of reality, given the enormous variation in vehicle types and conditions. Ideally, emission factors would be based on direct measurement of the exhaust from a relatively large sample of representative vehicles being driven under realistic conditions. Similarly, fleet composition and activity data (e.g., driving patterns and engine starts) should be based on on-the-ground surveys, using a protocol such as that developed by a team from UC Riverside and implemented in Pune, India (ISSRC 2004). However, since data collection requires extensive effort and cost, there are important knowledge gaps and significant uncertainties in emission inventories. In particular, there is a paucity of emissions and activity data for less developed countries. As mentioned earlier, it has long been recognized that one of the main challenges with determining emission factors for a given mode is accounting for the influence of ‘high-emitters’ (Beaton et al. 1995). This was demonstrated in a recent empirical study in Thailand, which determined emission factors for in-use diesel vehicles (Subramanian et al. 2009). That study showed that PM emission factors for ‘super-emitting’ vehicles 8.4 ± 1.9 g kg-1 (grams PM per kilogram of diesel-fuel) were almost four times that of ‘normal’ vehicles (2.2 ± 0.5 g kg-1). It is difficult to gather a representative sample of super-emitters because they comprise a relatively small fraction of the overall fleet. To address this problem, sample sizes in laboratory studies should be as large as possible – Subramanian and colleagues (2009) suggest as many as 25  16  vehicles – and techniques such as remote sensing or tunnel studies can also be effective (Ropkins et al. 2009). The activity term (Am,t) in equation 1.2 shows that vehicle activity is also a critical driver of emissions and thus reducing activity level is another potential objective for policies aiming to improve ambient AQ. Factors affecting activity include the numbers of vehicles, distance traveled and type of driving, all of which are heavily impacted by socio-economic factors. Some institutional mechanisms that have proved successful in controlling heavy-duty vehicle activity in urban areas include traffic management programs, congestion charges, vehicle-use limitations, and land-use planning, which lie within the domain of urban planning and the built environment (Pucher et al. 2005; Frank et al. 2000). Such policies aim to constrain vehicle activity in a jurisdiction while maintaining or improving mobility for goods and individuals. Examples include congestion charges, improvement of public transportation services, and promotion of non-motorized trip-making. Addressing rising activity levels is acknowledged to be an important part of the approach required to tackle the wide range of transportation-related problems facing developing Asian nations (Pucher et al. 2005); however an assessment of activity-related policies is beyond the scope of this thesis. It is important to recognize that per capita use of transportation services in developing countries is much lower than in more developed regions, so vehicle activity in Asia is anticipated to increase in the coming decades (Dargay et al. 2007). Thus policies or technologies that aim to control elevated emission factors – the topic of this dissertation – must play a strong role in efforts to improve air quality.  1.4. Emission control policies in Asia  Recognizing that emissions from transportation pose a serious health problem, especially in densely populated urban areas, Asian countries are following the lead of more developed nations by implementing policies to control emissions. The schematic in Figure 1.3 identifies the main factors that influence tailpipe emissions from motor vehicles, categorized into institutional, technological, behavioural and environmental factors. Policies can influence many of the institutional, technological and (to a lesser extent) behavioural factors. As introduced in the 17  previous section, the health impacts of motor vehicle emissions depend on the ambient concentration to which people are exposed, and the toxicity of the inhaled dose (this is known as the ‘exposure pathway’). Likewise, the climate impacts depend on the atmospheric concentration, lifetime, and global warming or cooling potential of each exhaust constituent. The main policy approaches are introduced and discussed in the following sub-sections.  Can be influenced by policy-makers  INSTITUTIONAL FACTORS  TECHNOLOGICAL FACTORS  •  Fuel quality regulations  •  Vehicle type (light-duty or heavy-duty)  •  Vehicle choice (for private vehicles)  •  Engine technology (sparkignition or compressionignition)  •  Activity (fuel/km per year)  •  Inspection & Maintenance programs •  Scrappage (based on age/technology)  •  Fuel type (gasoline, diesel, natural gas, biofuels...)  •  Activity-related regulations (e.g. congestion charges)  •  Emission aftertreatment (e.g. three-way catalyst)  •  Emission standards for new vehicles  ENVIRONMENTAL FACTORS •  Temperature, humidity, altitude  BEHAVIOURAL FACTORS  •  Maintenance practices •  Driving behaviour (e.g. aggressive acceleration) •  Disposal/replacement at end-of-life  Tailpipe Emissions  IMPACTS: 1. HEALTH:  Ambient Concentration  2. CLIMATE: Concentration & Lifetime  Exposure  Dose  Radiative Forcing  Health Effects Climate Impacts  Figure 1.3. Schematic of the factors that influence the mass (and composition) of tailpipe emissions from motor vehicles.  1.4.1. Emissions standards for new vehicles Perhaps the most well-established Indian policy effort to control emissions from vehicles has been the introduction of standards that require new vehicles to meet mandatory emission limits, modelled on increasingly stringent US and European regulations. The progressive lowering of emission limits over time is illustrated by PM standards for the diesel engines that power most HDVs (Fig. 1.4). Indian and Chinese regulations lag those in the US by approximately a decade. Emission limits for other pollutants from different modes (e.g., gaseous exhaust constituents 18  from LDVs) follow a similar pattern. The emission standards for new vehicles have forced auto manufacturers to add aftertreatment devices to the exhaust systems. As a result, modern vehicles produce only a fraction of the gaseous and particulate emissions of those from two decades ago, while fuel-specific performance has steadily increased (Oliver 2005). There is a delay between when emission standards for new vehicles are implemented and improvements in air-quality, since replacement of the existing (higher-emitting) fleet will take years or even decades (Calvert et al. 1993). Fleet turnover rates depend on the rate at which new vehicles are added, and the average age at which different classes of vehicle are retired. These data are highly uncertain in India because vehicles are only registered once when they are purchased and again at 15 years (rather than annually, which is the norm in many developed countries). Most trucks in India were likely manufactured before 2000, so in fact the Indian regulations (shown in Fig. 1.4) do not apply to most of the current fleet. That is not to say that India’s PM regulations are not effective, however: new vehicles with regulated emissions are replacing emissions from older vehicles with ‘uncontrolled’ engines, so the impact of mass emissions standards is likely to have a very substantial effect. Nonetheless, emissions control devices deteriorate as vehicles age, and engine malfunctions can cause tailpipe pollutant emissions to increase by orders of magnitude. Therefore it is desirable to identify means of controlling emissions from the in-use fleet in addition to implementing emissions standards for new vehicles. Approaches to achieve this include implementing inspection and maintenance programs, improving fuel quality, retrofitting vehicles with aftertreatment devices, and switching to alternative fuels.  19  PM emission emissionslimits regulations (g/kW-h) -1) PM (g kW-h  0.5  US HD Trucks US HD Urban Buses EU HD Vehicles  0.4  India HD Vehicles China HD Vehicles  0.3  0.2  0.1  0.0 1985  1990  1995  2000  2005  2010  2015  Year  Figure 1.4. Particulate matter (PM) mass emissions standards for new heavy-duty vehicles in different regions over time. Emissions standards in developing countries lag those in the developed world. Note that some cities in India/China (e.g., Beijing, Delhi and other major Indian cities) must meet the standards earlier than the nationwide requirements.  1.4.2. Exhaust aftertreatment devices for new and in-use vehicles The use of catalytic emission-control devices in light- and heavy-duty vehicles has resulted in dramatic reductions in motor-vehicle emissions since they were introduced in the early 1970s. In the US, the motivation was the 1970 Clean Air Act, which forced auto-manufacturers to look beyond engine modification as means of reducing tailpipe emissions to meet newer, more stringent, limits. Twigg (2007) provides a comprehensive review of the development and implementation of catalytic emission-control technologies. As mentioned earlier, catalysts are prone to ‘poisoning’ by fuel contaminants such as lead and sulphur, so strict control of fuel quality is a prerequisite for successful introduction of catalytic devices. Most 4-stroke SI engines now use three-way catalysts (TWCs), so-called because they simultaneously convert the three main gaseous pollutants: CO and THC are oxidized to CO2 and H2O while NOX is chemically reduced to N2 and O2. Emissions from small-displacement engines, such as those found in motorcycles, are harder to control than larger, multi-cylinder engines: in an assessment of the real-world performance of TWCs on motorcycles with Euro-3 certification, it was shown that although their emissions were significantly lower than earlier certification classes, they exceeded the emission levels of most modern cars (Alvarez et al. 2009). Although TWCs are not 20  specifically designed to reduce PM, there is strong evidence that uncontrolled gasoline-fueled vehicles with 4-stroke engines emit 3-10 times more PM than vehicles fitted with TWCs (Maricq et al. 2002a). Catalytic control of CO and HC emissions from 2-stroke engines is usually achieved with oxidation catalysts, which can also reduce PM by up to ~50% (Alander et al. 2005; Volckens et al. 2007). In compression-ignition diesel engines, aftertreatment devices are very important for control of particulate pollutants (Lloyd and Cackette 2001; Maricq et al. 2002b), and they are typically comprised of two stages. First, a diesel oxidation catalyst (DOC) oxidizes CO and THC to CO2, and NO to NO2. The NO2 produced in the DOC is a powerful oxidant, which is important in the second stage of the process. After the DOC, a catalyzed diesel particulate filter (DPF) physically traps engine-generated PM and oxidizes it to CO2 and water. DOCs alone are less expensive and more robust than particulate traps, but they are also less effective at reducing emitted PM mass. In an experiment where DOCs were retrofitted to buses operating on ‘low-sulphur’ diesel (<500ppm), PM was reduced by 30-75% (Bose and Sundar 2005). While DOCs can reduce PM from diesel engines by about half, when used with a DPF the PM-reduction efficiency increases to >90% (Frank et al. 2004). Also, it is important to recognize that DOCs reduce the organic fraction of PM, but has little effect on EC. Control of NOX emissions is another challenge for compression-ignition engines. Strategies to reduce NOX emissions include lean-NOX traps and urea-based selective catalytic reduction (Johnson 2004). There are a number of examples where HDVs have been successfully retrofitted with emissions control devices in the US and worldwide (MECA 2009), and recent modeling studies suggest that this approach merits further consideration (Stasko and Gao 2010; Millstein and Harley 2010). A disadvantage of all catalytic emission control devices is that they are expensive, and durability may be an issue where fuel quality is inconsistent. In addition, catalytic emission control devices tend to reduce fuel efficiency because they increase back-pressure in the exhaust system. This increases fuel consumption and CO2 emissions.  1.4.3. Fuel quality regulations The reduction of contaminants in petroleum-based fuels is important for emissions control, and mitigation of adverse health effects. An exhaustive discussion of fuel quality regulations is not 21  possible here, but some key developments are identified and discussed briefly. The removal of lead, which is a potent neurotoxin, from gasoline has been described as one of the greatest public health achievements of the 20th century (Bridbord and Hanson 2009). There is also evidence that combustion of unleaded gasoline produces less HC, CO and PM emissions (Yuan et al. 2000). More recently, there have been efforts to limit the content of the carcinogen benzene in gasoline, since public exposure to this chemical is primarily from mobile sources (Smith 2010). The amount of SO2 emitted from an engine is directly proportional to fuel sulphur content (SO2 is a precursor to sulphate aerosol). More importantly, low-sulphur fuel is a ‘technology enabler’, which allows the use of advanced emissions control devices such as the TWC for SI engines, and the DOC and DPF for CI engines. Most developed countries have regulated sulphur levels in gasoline and diesel fuel for on-road motor vehicles to levels that are orders of magnitude lower than past standards, so that emission control devices can be used. In North America and Europe, fuel sulphur content is now less than 15ppm. Many rapidly-industrializing Asian countries are following the trend towards lower sulphur concentrations in on-road vehicle fuel, though they are up to a decade behind more-developed countries. Within India, for example, diesel fuel with sulphur content of less than 350ppm is available in selected urban areas (and may be reduced to 50ppm as early as end-2010), compared with 500ppm nationwide (UNEP 2009). Asian countries with diesel sulphur levels of greater than 2000ppm include Afghanistan, Bangladesh, Indonesia, Laos, Mongolia and Pakistan. There is evidence that contamination of diesel and gasoline with cheaper hydrocarbons such as kerosene (also known as fuel adulteration) is a widespread problem in India (CPCB 2003) and probably in other developing nations. Fuel adulteration can have substantial negative impacts on engine emissions and performance. Therefore systematic fuel testing (e.g., in vehicles or fuel distribution centres) to detect possible adulteration of diesel has potential to substantially reduce vehicle emissions in certain cases.  1.4.4. Emission control policies for the in-use fleet Older vehicles are less likely to have modern emission control devices, and it is more likely that their engines will be worn or in a poor state of repair. As a result, vehicle age (in terms of years or kilometres traveled) is positively correlated with higher pollutant emissions (Beaton et al. 22  1995). Newer vehicles also tend to be more efficient and thus have lower GHG emissions. Policies to scrap older vehicles in a region can be an effective means of achieving co-benefits, in terms of both PM and GHG emission reductions. In New Delhi, a comprehensive program to scrap commercial vehicles older than 15 years was phased in over a six-year period (1996-2002) as part of a suite of measures to reduce transportation emissions (Jalihal and Reddy 2006). However, a major problem with age-based vehicle scrappage programs is that they are likely to disproportionately impact less-wealthy vehicle owners and operators. It would therefore be regressive from an equity perspective to require large-scale replacement of in-use vehicles before the end of their useful life, especially in resource-constrained economies. In addition, it is well known that vehicle age is not the only predictor for vehicle emissions. In addition to vehicle age, it has been consistently found that the majority of emissions are from a small fraction of malfunctioning vehicles (Beaton et al. 1995; Mazzoleni et al. 2007). When implemented correctly, inspection and maintenance (I/M) programs can be an effective way of identifying those high-emitters. I/M programs require that vehicle (or fleet) owners have their vehicles’ emissions checked at regular intervals4. If emissions measurements do not meet the regulatory limit for a given vehicle’s category, the program then either levies a fine or can embargo vehicle registration pending engine maintenance and retesting. Mandatory vehicle retirement may be considered as an option only once repair of high-emitting vehicles is ruled out as an option. Internationally, most existing I/M programs are designed to control gaseous emissions from light-duty vehicles (St. Denis et al. 2005), but they have also been used successfully with heavyduty fleets (Van Houtte and Niemeier 2008). A recent review of the effectiveness of I/M programs for monitoring PM emissions from heavy-duty vehicles identified 19 programs in Asia and Latin America (Van Houtte and Niemeier 2008). One reason for the lack of widespread HDV emission testing is that the instrumentation required to accurately measure PM from vehicles is more complex and expensive than for gaseous pollutants. The most comprehensive study of the effectiveness of HDV I/M programs found an average reduction in PM emission factor of more than 40% with repairs on vehicles exhibiting visible smoke emissions; average per-vehicle repair costs in the study were approximately $1000 (McCormick et al. 2003). The overall cost-effectiveness of such an intervention would depend on the prevalence and usage of 4  Remote sensing or computer-assisted inspections (that rely on the on-board diagnostic systems present in most vehicles) can be used to screen large numbers of vehicles and identify candidates for emission measurement (Eisinger and Wathern 2008); this approach could improve the cost-effectiveness of an I/M program.  23  high-emitting vehicles along with the programmatic costs. Improved fuel economy (and corresponding reduced GHG emissions) can be a co-benefit of I/M programs, because bettermaintained engines convert fuel into mechanical energy more efficiently. In theory, I/M programs ensure that all vehicles operating in a given region emit less than the inuse emission standard set by regulators. However, due to the paucity of actual emissions measurements, conducting assessments on a program-by-program basis has been difficult. In general, I/M programs are susceptible to inaccuracy, corruption and high administrative costs, and must be carefully designed to be successful (Hausker 2004). The Indian ‘Pollution Under Control’ (PUC) program is an example of such problems (Rogers 2002). A detailed evaluation and critique of I/M programs, including case-studies of Mexico and New Delhi, was performed by the US Agency for International Development (Hausker 2004), and offers the following criteria for a successful and worthwhile program: •  An I/M program should conduct inspections using centralized ‘test-only’ facilities, not decentralized ‘test-and-repair’ operations.  •  Government should set the policy framework and provide overall management of the I/M program while private contractors perform the actual inspections.  •  Policymakers should exert strong oversight and institute a quality assurance (QA) program for the I/M program.  •  Policymakers should implement I/M programs in a phased approach that allows learning, adaptation, and capacity building along the way.  1.4.5. Alternative fuels Use of alternative fuels in place of gasoline and diesel has been advocated as a means of reducing emissions and improving fuel security. The alternative fuel options for SI engines include compressed natural gas (CNG), liquefied petroleum gas (LPG), and ethanol. Biodiesel can be used directly in CI engines, but use of other alternative fuels in HDVs – such as CNG – normally requires major engine modification to include a spark-ignition system (McTaggartCowan et al. 2006). LPG is gaseous at atmospheric pressure but can be more readily compressed than natural gas, which makes it easier to store on-board vehicles. However, it still requires a dedicated fueling system (compared to liquid biofuels, which can be used in conventional engines in many cases). LPG represents only a small fraction (around 3%) of current 24  hydrocarbon fuel production, so its main potential is as a niche fuel for urban vehicles, such as buses or delivery fleets. Biofuels have been proposed as a promising way of reducing the carbon intensity of fuel (e.g., Hill et al. 2006). The impacts of biofuel use on pollutant emissions are uncertain and depend on a range of factors (e.g., fuel quality, engine compatibility). Biodiesel may reduce PM emission by a small amount: some studies found use of a B20 blend (20% biodiesel in conventional diesel) resulted in ~10% lower PM emissions (USEPA 2002). However, a more recent study found PM increased by up to a factor of two, potentially due to variable biodiesel quality (Mazzoleni et al. 2007). Ethanol use in SI engines could also be problematic: a modeling study estimated that widespread use of E85 (85% ethanol, 15% gasoline) in the US could lead to a 4% increase in ozone-related mortality and morbidity, and increase emissions of some air toxics such as aldehydes (Jacobson 2007). Additionally, there is concern that land-use change related to growing biofuel crops might lead to a net increase in GHG emissions (Fargione et al. 2008; Searchinger et al. 2008). Due to these significant uncertainties, a detailed assessment of the health and climate impacts of biofuels is outside the scope of this thesis. In this thesis, I examine the impact of switching from conventional liquid fossil fuels to CNG. Natural gas is a far more abundant fossil fuel than LPG, and compression and storage technologies have improved greatly in recent years, reducing CNG’s cost. Natural gas is cleanerburning than gasoline or diesel because methane (the main component of CNG) lacks carboncarbon bonds and hence is less likely to form PM or carcinogenic polycyclic aromatic hydrocarbons (Warnatz et al. 1999). Purpose-built CNG and LPG heavy-duty engines are available, but retrofitting existing diesel engines offers a lower-cost alternative. This approach has been taken in a number of major South Asian cities, including New Delhi, Mumbai and Dhaka. Although use of CNG in SI engines is relatively straightforward, converting a CI engine to run on CNG is more challenging (and expensive) because an appropriate ignition system must be added. Heavy-duty natural gas engines are normally less fuel-efficient than the diesel engines they replace, but their CO2 emissions per kilometre may be approximately equivalent because CNG has lower fuel carbon content (Frailey and Norton 2000).  25  CNG fueling offers the potential for dramatic reductions in PM emissions from heavy-duty trucks and buses, by one to two orders of magnitude (Rabl 2002; Hesterberg et al. 2008). It has been demonstrated that installing oxidation catalysts on CNG fueled vehicles further reduces PM emissions (Ayala et al. 2003). One of the most significant fuel-switching examples has been in Delhi, where the entire public transportation fleet – buses, taxis and auto-rickshaws (threewheeled motorcycle taxis) – were converted to run entirely on natural gas (Mehta 2001). The switch was mandated by a Supreme Court of India directive (in response to a public petition), which required that the Delhi Government tackle the problem of air pollution from publictransport vehicles. The apparent benefits of CNG must be kept in perspective; modern engines with functioning catalytic emission control devices can have similar PM emission factors to CNG vehicles. A meta-analysis by Hesterberg et al. (2008) found that PM emissions from CNG vehicles were not statistically different from modern diesel vehicles with regenerating DPFs. A potential advantage of CNG fuel is that engine-out emissions are generally much lower than untreated emissions from diesel- or gasoline-fueled engines, so the consequence of a malfunctioning aftertreatment device is less severe. A potential negative impact of converting vehicles to operate on CNG is increased methane emissions during vehicle fueling and use, since methane is a potent GHG. Life-cycle impact assessment methodologies can be an effective way of evaluating the impacts of a policy, and assist policy-makers make informed environmental decisions (e.g. Rogers and Seager 2006). In this thesis I describe an integrated assessment of the Delhi fuel-switching policy, where around 90,000 vehicles were converted to run on CNG (Chapter 2). The increase in methane emissions was estimated to have offset much of the climate benefit due to diesel PM reductions.  1.4.6. New powertrain technologies Reduced fuel use via improved vehicle efficiency may offer significant climate and health cobenefits (Smith and Haigler 2008). Alternative vehicle designs using advanced lightweight materials and non-conventional powertrains (e.g., hybrid-electric) are proposed to be a viable means of achieving this goal (Zhao 2006). New powertrain technologies also tend to use stateof-the-art emissions control technologies. At this time, such technologies are significantly more  26  expensive than conventional powertrain technologies and are thus unlikely to be a viable approach for GHG and PM mitigation in Asia in the next decade.  1.5. Framework to evaluate co-impacts  The framework proposed here is an integrated approach to the analysis of pollution-reduction strategies that accounts for both the health- and climate-damaging co-impacts of pollution, as recommended by Dowlatabadi (2007). Variants of this framework are used throughout this thesis. Using such a co-impacts approach to emissions reduction, strategies that address both the local health impacts and the regional- and global-scale climate impacts of energy-use decisions can be formulated and evaluated. Active mitigation of health-harming pollutant emissions should be done in such a way that accounts for their climate impacts, and vice-versa. Other considerations, such as the cost, technical and institutional feasibilities and social acceptability of mitigation approaches must also enter into the decision-making framework. A quantitative assessment of these other dimensions is difficult because, in many cases, they are unknown and/or highly context-dependent and thus cannot be generalized. However, discussions of technological feasibility and socio-economic issues have been included in the thesis wherever possible. A simple conceptual model that qualitatively describes the health/climate co-impacts framework is shown in Figure 1.5. Any policy or action that is taken to address PM emissions will cause a shift from the origin – the status quo – into one of the four quadrants in the health/climate domain (Fig. 1.5.A). It should be clear that co-benefits are not guaranteed. A policy that results in lower PM emissions (i.e., resulting in a health benefit) but that also causes a net increase in the GWC (i.e., a climate dis-benefit) would cause a shift to point ‘X’, as shown by the dashed arrow. A decision analysis may reveal that such a trade-off may be acceptable, because the health benefits far outweigh the climate dis-benefits. The critical point is that all of these impacts are recognized and included in the decision-making process. Such decisions are highly context dependent, as relative valuation of different impacts varies with location and stakeholder. For example, health benefits (quantified directly or in economic terms) accrued in developing Asian countries should be compared to mitigation costs, while also considering potential resource  27  transfers from climate offsets markets (including the IPCC’s Clean Development Mechanism and other voluntary markets) in more-developed countries. The health and climate impacts of the various emissions control options presented in the previous section are presented qualitatively in Figure 1.5.B. In principle, reducing emissions from vehicles should have benefits from both air quality and climate perspectives (Bell et al. 2008; Smith and Haigler 2008). In practice, how reductions of health-relevant emissions are achieved might determine the magnitude (and in some cases the sign) of climate benefits. This graphical representation of the co-impacts of different emission mitigation options also allows the uncertainties associated with each option to be displayed. For example, the oval displaying the possible outcomes for a CNG fuel switch show that such an approach could have either climate benefits or dis-benefits, depending on how such a program is implemented and the relative changes in methane, EC and other climate forcing emissions. Chapter 2 in this dissertation explores this issue in more detail, using Delhi as a case-study. There is also significant uncertainty in the level of health benefits due to a CNG switch, because PM emission reductions and the health impacts of PM are uncertain. Chapter 4 addresses this knowledge gap for light-duty vehicles, by measuring gaseous and particulate pollutants from in-use autorickshaws with different fuel/engine combinations. Other mitigation options similarly show a range of potential outcomes in both climate- and health-related impacts. More detailed analysis can reduce the extent of these uncertainty ovals. However, the key point here is that the importance of uncertainty must be appreciated during decision-making processes.  28  Inspection/Maintenance programs reduce emissions from in-use vehicles, but require significant government investment (financial and human resources) to be successful  B  A Climate Benefit  UK Vehicle Excise Tax that inadvertently encouraged LDVs to shift from gasoline to diesel, which emitted higher PM  Co-benefits Quadrant Q4  I/M Program  Q1  Health Dis-benefit Q3  Climate Benefit  Q2  Climate Dis-benefit  X  Health Benefit  Health Dis-benefit  Clean Fuel  CNG CNG-fuelled vehicles can emit less PM, but they also emit significant methane, which is a greenhouse gas  Climate Dis-benefit  Exhaust After-treatment  Health Benefit  Three-way catalysts and diesel particulate filters could reduce engine efficiency and hence increases CO2  Improving fuel quality (e.g. reducing sulphur or benzene content) can reduce emissions, and it is also a prerequisite for exhaust after-treatment devices  Figure 1.5. Conceptual diagram of co-impacts framework for emissions control policies – health (PM-reduction) or climate (GWC) policies can all be located in this space. A: Quadrant 1 (Q1) is the only space that has both climate and health co-benefits. The shift to point ‘X’ (dashed arrow) in Q2 represents an emissions-related policy that has a health benefit but a climate dis-benefit. ‘Health benefit’ means a cut in PM emissions (leading to reduced ambient PM concentrations and less human exposure); ‘climate benefit’ means a reduction of CO2 equivalent emissions (both GHGs and PM). B: The hypothetical co-impacts of various emission-reduction options. The size of the ovals represents the authors’ subjective uncertainty space for each policy/technology option. Considering other aspects of decisions (e.g., cost and feasibility) would add further ‘dimensions’ to this framework (not shown on this figure).  Any climate policies that would have significant health dis-benefits would be highly undesirable, and pursuing such policies could be considered unethical. However, there are real cases of past and current climate policies that have had adverse health outcomes. For example, a CO2-based vehicle tax in the UK resulted in an increase in the number of light-duty diesel vehicles, leading to higher fleet aggregate PM emissions and consequent adverse health impacts (Mazzi and Dowlatabadi 2007). This outcome occurred because the CO2-control policy was not harmonized with more stringent PM controls for diesel vehicles; in essence, only the climate axes in Figure 1.5 were considered. Had a co-impacts analysis been conducted prior to implementation, negative health impacts might have been avoided. Such interactions between the health and  29  climate domains indicate a clear need for new, integrated approaches to better quantify and understand the co-impacts of PM reduction strategies.  1.5.1. Addressing critical knowledge gaps From an air quality perspective, policies in Asia have had varying degrees of effectiveness. A good example of an evidence-based emissions-control policy in Asia is the ongoing Developing Integrated Emissions Strategies for Existing Land-transport (DIESEL) project, conducted by the Bangkok Pollution Control Department and supported by the World Bank (DIESEL 2008). On the other hand, an unsystematic approach to air quality management in New Delhi has lead to an ineffective policy – in the case of the ‘Pollution under Control’ inspection and maintenance program (Rogers, 2002) – and a questionably effective policy – in the case of the CNG fuel switch for public transit (Kandlikar 2007; Reynolds and Kandlikar 2008). Emission controls implemented in the United States and Europe have been more successful, driven by welldesigned policies and informed by evidence-based decision-making. Urban air quality has steadily improved in North American cities despite consistent increases in activity levels (USEPA 2004). The same is also true for many Asian cities, and it has happened over a much shorter time period. Policies should be implemented because there is sound evidence that they will be the most cost-effective means of reducing health-relevant emissions (primarily PM) without adverse co-impacts. Better data is essential if we are to understand how best to reduce and control emissions without negative co-impacts, so governments should support initiatives to fill this knowledge gap. However, research programs to measure emissions and understand their determinants can be very costly, so they must be carefully designed to ensure that the cost of collecting the information does not exceed the benefits that are expected to accrue from the study. Given the huge diversity in transportation types and uses across Asia, identification of programs that can be implemented effectively at different scales (e.g., urban, regional, national) could streamline the adoption of such measures. Improving knowledge transfer channels between institutions (both at a national as well as international level) will be essential to ensure that Asian countries are making effective decisions for air pollution and climate change mitigation.  30  1.6. Overview of dissertation  This is a ‘manuscript-based’ dissertation. It is comprised of this introductory chapter, four analytical chapters, and a concluding chapter. The four research chapters were written as standalone manuscripts for publication in the peer-reviewed academic literature (I have indicated at the beginning of each chapter if the manuscript is undergoing review, or has been published). Consequently, and unlike a ‘monograph-style’ dissertation, each chapter contains a contextspecific introduction, description of methods, presentation of results, and a discussion/conclusion section. Some repetition across chapters has been necessary so that each piece is a complete description of the research in question. In order to meet individual journal requirements for manuscript length and format, ‘supporting online information’ documents were created for chapters 2-4 (including details of methods). These have not been included within the chapters, but instead are added to the end of the dissertation, as appendices. Chapter 1 (this introductory chapter) provides a statement of the research questions that have driven this thesis, and gives background on the importance of transportation as a source of emissions that causes local/regional air pollution and global climate change. The motivations for conducting research in a developing-country context have been elaborated, and the various policy and technology options for reducing vehicle emissions were described. A simple analytical framework has been presented for assessing the co-impacts of emissions-reduction policies. This framework sets the stage for the subsequent chapters, which examine specific options in more detail. A key hypothesis generated by this framing process is that not all emissions control approaches have comparable potential: air quality and climate co-benefits are not automatic, and that there may be adverse economic, social or environmental ‘co-impacts’ if emission-reduction policies are not carefully designed. Chapter 2 assesses the climate impacts of an air quality policy in Delhi, whereby approximately 90,000 public transport vehicles (diesel buses, gasoline taxis and gasoline auto-rickshaws) were required to switch their fuel to natural gas, in an attempt to reduce urban air pollution. Using an integrated assessment approach, I found that methane and black carbon emissions were critical contributors to the change in GWC. In New Delhi, the switch to natural gas resulted in a 30% increase in GWC when the impact of aerosols was not considered. However, when aerosol 31  emissions are taken into account in our model, the net effect of the switch is estimated to be about a 10% reduction in GWC. Implications for similar fuel-switching efforts in other rapidly industrializing countries are outlined. Chapter 3 investigates the determinants of PM and GHG emissions from CNG-fueled autorickshaws in Delhi. Auto-rickshaws are an important mode of transportation for a large segment of the population in many developing Asian cities. 381 drivers of CNG-fueled auto-rickshaws were interviewed using a structured survey. Quantitative, policy-relevant information about the characteristics of the fleet were determined, including activity factors (distance traveled per day), fuel consumption and CO2 emissions. Delhi auto-rickshaws were found to travel approximately 150 km per day. Vehicles with 4-stroke engines (which made up 90% of the fleet) were estimated to have about 20% lower fuel consumption and CO2 emissions than auto-rickshaws with 2-stroke engines. An inspection of each vehicle was conducted (to check for oil residue in the tailpipe and visible smoke at engine start-up) and vehicles were consequently classified as either low- or high-PM emitters. This method was checked against a subset of vehicles, for which PM emissions factors were measured during chassis-dynamometer testing. Statistical analysis indicated that vehicles with 2-stroke engines had a much higher likelihood of being categorized as high-PM emitters than those with 4-stroke engines. Within the group of 4-stroke vehicles, age was a highly significant predictor of high-emitters. The results of this study suggest that this simple observational procedure could be used to rapidly identify potential PM ‘superemitters’ for further testing and repair. Chapter 4 reports on a laboratory study to quantify the climate- and health-relevant emissions from in-use auto-rickshaws. Chassis dynamometer emission testing was conducted on 30 in-use auto-rickshaws to determine fuel-based emission factors from CNG-fueled spark-ignition engines, and compared to gasoline-fueling. Emission factors were determined for gaseous pollutants (CO2, CH4, NOX, THC, and CO) and fine particulate matter (PM2.5). Inter-vehicle variability was high, and for most pollutants there was no significant difference between ‘old’ (1998-2001) and ‘new’ (2007-2009) age-groups within a given fuel-technology class. The mean fuel-based PM2.5 emission factor for the CNG 2-stroke engines (14.2 g kg-1) was almost thirty times higher than for the CNG 4-stroke engines (0.5 g kg-1) and about twelve times higher than for the gasoline 4-stroke engines (1.2 g kg-1). Global warming commitment associated with 32  emissions from the CNG 2-stroke engines was more than twice that from the 4-stroke engines, due mostly to CH4 emissions. This study confirmed that comprehensive measurements and emission data should drive policy interventions rather than assumptions about the impacts of clean fuels. Chapter 5 places the findings from chapters 2, 3 and 4 into the analytical framework presented in the introduction, and provides estimates of the impacts of a range of policy options on emission factors for light-duty vehicles. AQ and climate-relevant emission factors were compared for auto-rickshaws with 2-stroke (2S) and 4-stroke (4S) engines, and gasoline/petrol (PET) and compressed natural gas (CNG) fuel. The effectiveness of various emissions-control policies for these groups were then modeled and assessed, including phasing out 2S engines, switching to CNG, an age-based scrappage scheme and four different types of inspection and maintenance programs. The data and modeling results reinforce the recommendation that it is essential to favor robustly ‘clean’ technologies, such as the CNG-4S engines in Delhi’s autorickshaw fleet, if health and climate co-benefits are to be realized. Therefore high-emitting 2S engines should be phased out wherever possible. Inspection and maintenance programs based on dynamometer measurements of gaseous emissions are not very effective at reducing fleet PM emissions. However, simple inspections for visible smoke can identify high PM-emitting 2S vehicles, which may be necessary to control the impact of remaining 2S vehicles in a fleet.  33  1.7. References Alander, T.; Antikainen, E.; Raunemaa, T.; Elonen, E.; Rautiola, A.; Torkkell, K. Particle emissions from a small two-stroke engine: Effects of fuel, lubricating oil, and exhaust aftertreatment on particle characteristics. Aerosol Sci. Tech. 2005, 39 (2), 151-161. Alvarez, R.; Weilenmann, M.; Favez, J. Assessing the real-world performance of modern pollutant abatement systems on motorcycles. Atmos. Environ. 2009, 43 (8), 1503-1509. Anderson, H. R.; Atkinson, R.; Chen, B.; Cohen, A.; Greenbaum, D.; Hedley, A. J.; Huang, W.; Pande, J.; Pope, C. A., III; Smith, K. Health effects of outdoor air pollution in developing countries of Asia: A literature review. Health Effects Institute: Boston, MA, 2004. Ayala, A.; Gebel, M.; Okamoto, R. A.; Rieger, P. L.; Kado, N.; Cotter, C.; Verma, N. Oxidation catalyst effect on CNG transit bus emissions. SAE Tech. Pap. 2003, 2003-01-1900. Baidya, S.; Borken-Kleefeld, J. Atmospheric emissions from road transportation in India. Energy Policy. 2009, 37 (10), 3812-3822. Ban-Weiss, G. A.; Lunden, M. M.; Kirchstetter, T. W.; Harley, R. A. Measurement of black carbon and particle number emission factors from individual heavy-duty trucks. Environ. Sci. Technol. 2009, 43 (5), 1419-1424. Beaton, S. P.; Bishop, G.A.; Zhang, Y.; Stedman, D. H.; Ashbaugh, L. L.; Lawson, D. R. Onroad vehicle emissions: Regulations, costs, and benefits. Science, 1995, 268 (5213), 991-993. Bell, M.; Davis, D.; Cifuentes, L.; Krupnick, A.; Morgenstern, R.; Thurston, G. Ancillary human health benefits of improved air quality resulting from climate change mitigation. Environ. Health. 2008, 7 (1); DOI: 10.1186/1476-069X-7-41. Bennett, D. H.; McKone, T. E.; Evans, J. S.; Nazaroff, W. W.; Margni, M. D.; Jolliet, O.; Smith, K. R. Defining intake fraction. Environ. Sci. Technol. 2002, 36 (9), 206A-211A. Bond, T. Can warming particles enter global climate discussions? Environ. Res. Lett. 2007, 2 (4), 045030. Bond, T.; Streets, D. G.; Yarber, K. F.; Nelson, S. M.; Woo, J. H.; Klimont, Z. A technologybased global inventory of black and organic carbon emissions from combustion. J. Geophys. Res. 2004, 109 (D14203). Bond, T.; Sun, H. Can reducing black carbon emissions counteract global warming? Environ. Sci. Technol. 2005, 39 (16), 5921-5926. Bose, R.; Sundar, S. Emissions test results from diesel buses, with and without oxidation-catalyst and regenerating particle-trap, and CNG buses with three-way catalyst in India. SAE Tech. Pap. 2005, 2005-01-0477. 34  Bridbord, K.; Hanson, D. A personal perspective on the initial federal health-based regulation to remove lead from gasoline. Environ. Health Persp. 2009, 117 (8), 1195-1201. Calvert, J. G.; Heywood, J. B.; Sawyer, R. F.; Seinfeld, J. H. Achieving Acceptable Air Quality: Some Reflections on Controlling Vehicle Emissions. Science. 1993, 261 (5117), 37-45. Campbell-Lendrum, D.; Corvalán, C. Climate change and developing-country cities: Implications for environmental health and equity. J. Urban Health. 2007, 84 (3 Suppl), i109117. Chan, C.; Yao, X. Air pollution in mega cities in China. Atmos. Environ. 2008, 42 (1), 1-42. Chowdhury, Z.; Zheng, M.; Schauer, J. J.; Sheesley, R. J.; Salmon, L. G.; Cass, G. R.; Russell, A. G. Speciation of ambient fine organic carbon particles and source apportionment of PM2.5 in Indian cities. J. Geophys. Res. 2007, 112 (D15303). Cocker, D. R., III; Shah, S. D.; Johnson, K. C.; Zhu, X.; Miller, J. W.; Norbeck, J. M. Development and application of a mobile laboratory for measuring emissions from diesel engines. 2. Sampling for toxics and particulate matter. Environ. Sci. Technol. 2004, 38 (24), 6809-6816. Clark, N. N.; Kern, J. M.; Atkinson, C. M.; Nine, R. D. Factors affecting heavy-duty diesel vehicle emissions. J. Air Waste Manag. Assoc. 2002, 52 (1), 84-94. CPCB. Transportation fuel adulteration. Central Pollution Control Board: Delhi, India; 2003; http://cpcbenvis.nic.in/newsletter/fueladultration/content0703.htm. Cropper, M. L.; Simon, N. B.; Alberini, A.; Arora, S.; Sharma, P. K. The health benefits of air pollution control in Delhi. Amer. J. Agri. Econ. 1997, 79 (5), 1625-1629. Dargay, J.; Gately, D.; Sommer, M. Vehicle ownership and income growth, worldwide: 19602030. Energy J. 2007, 28 (4), 143-170. Delfino, R. J.; Sioutas, C.; Malik, S. Potential role of ultrafine particles in associations between airborne particle mass and cardiovascular health. Environ. Health Persp. 2005, 113 (8), 934-946. DIESEL. Developing Integrated Emissions Strategies for Existing Land-Transport (DIESEL) Program for Bangkok. UNDP/World Bank Energy Management Assistance Programme: New York, NY, 2005; www.cleanairnet.org/caiasia/1412/article-48845.html. Dieselnet Website. Diesel Exhaust Emissions Standards; www.dieselnet.com/standards. Dowlatabadi, H. On integration of policies for climate and global change. Mitig. Adapt. Strat. Global Change. 2007, 12 (5), 651-663. Fargione, J.; Hill, J.; Tilman, D.; Polasky, S.; Hawthorne, P. Land clearing and the biofuel carbon debt. Science. 2008, 319 (5867), 1235-1238.  35  Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.; Haywood, J.; Lean, J.; Lowe, D.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Dorland, R. V. Changes in atmospheric constituents and in radiative forcing, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon, S. et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, NY, 2007. Frailey, M. R.; Norton, P. An evaluation of natural gas versus diesel in medium-duty buses. SAE Tech. Pap. 2000, 2000-01-2822. Frank, B. P.; Tang, S.; Lanni, T.; Rideout, G.; Beregszaszy, C.; Meyer, N.; Chatterjee, S.; Conway, R.; Lowell, D.; Bush, C.; Evans, J. A study of the effects of fuel type and emission control systems on regulated gaseous emissions from heavy-duty diesel engines. SAE Tech. Pap. 2004, 2004-01-1085. Frank, L. D.; Stone, B.; Bachman, W. Linking land use with household vehicle emissions in the central Puget Sound: Methodological framework and findings. Transp. Res. D – Tr. E. 2000, 5, 173-196. Gillies, J.; Gertler, A. Comparison and evaluation of chemically speciated mobile source PM2.5 particulate matter profiles. J. Air Waste Manag. Assoc. 2000, 50 (8), 1459-1480. Grieshop, A. P.; Reynolds, C. C. O.; Kandlikar, M.; Dowlatabadi, H. A black-carbon mitigation wedge. Nature Geo. 2009, 2 (8), 533-534. Gurjar, B.; Butler, T. M.; Lawrence, M. G.; Lelieveld, J. Evaluation of emissions and air quality in megacities. Atmos. Environ. 2008, 42 (7), 1593-1606. Guttikunda. S. K.; Tang, Y. H.; Carmichael, G. R.; G Kurata, G.; Pan, L.; Streets, D. G.; Woo, J. H.; Thongboonchoo, N.; Fried, A. Impacts of Asian megacity emissions on regional air quality during spring 2001. J. Geophys. Res. 2005, 110 (D20), D20301. Hansen, J.; Nazarenko, L. Soot climate forcing via snow and ice albedos. PNAS. 2004, 101 (2), 423-428. Hansen, J.; Sato, M.; Kharecha, P.; Russell, G.; Lea, D. W.; Siddall, M. Climate change and trace gases. Philos. Trans. R. Soc. London, Ser. A. 2007, 365 (1856), 1925–1954. Hausker, K. Vehicle inspection and maintenance programs: International experience and best practices. US Agency for International Development, Office of Energy and Information Technology: Washington, DC, 2004: http://pdf.usaid.gov/pdf_docs/PNADB317.pdf. HEI. Traffic-Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Special Report 17; Health Effects Institute: Boston, MA, 2010. Hesterberg, T. W.; Lapin, C. A; Bunn W. B. A comparison of emissions from vehicles fueled with diesel or compressed natural gas. Environ. Sci. Technol. 2008, 42 (17), 6437-6445.  36  Heywood, J. B. Internal Combustion Engine Fundamentals. McGraw-Hill: New York, NY, 1988. Hill, J.; Nelson, E.; Tilman, D.; Polasky, S.; Tiffany, D. Environmental, economic, and energetic costs and benefits of biodiesel and ethanol biofuels. PNAS. 2006, 103 (30), 11206-11210. Hopke, P. K.; Cohen, D. D.; Begum, B. A.; Biswas, S. K.; Ni, B.; Pandit, G. G.; Santoso, M.; Chung, Y.; Davy, P.; Markwitz, A.; Waheed, S.; Siddique, N.; Santos, F. L.; Pabroa, P. C. B.; Seneviratne, M. C. S.; Wimolwattanapun, W.; Bunprapob, S.; Vuong, T. B.; Hien, P. D.; Markowicz, A. Urban air quality in the Asian region. Sci. Tot. Environ. 2008, 404 (1), 103-112. IEA. World Energy Outlook 2006. International Energy Agency: Paris, France, 2006. IPCC. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, 2007. ISSRC. Pune vehicle activity study (March 9-March 22 2003). International Sustainable Systems Research Centre: La Habra, CA, 2004; www.issrc.org. Iyer, N.V. Role of three-wheeled vehicles in urban transportation in South Asia. Smart Urban Transp. 2003, 2 (2). Jacobson, M. Z. Effects of ethanol (E85) versus gasoline vehicles on cancer and mortality in the United States. Environ. Sci. Technol. 2007, 41 (11), 4150-4157. Jalihal, S. A.; Reddy, T. S. Assessment of the impact of improvement measures on air quality: Case study of Delhi. J. Transp. Eng. 2006, 132 (6), 482-488. Johnson, T. V. Diesel Emission Control Technology: 2003 in Review. SAE Tech. Pap. 2004, 2004-01-0070. Kahn Ribeiro, S.; Kobayashi, S.; Beuthe, M.; Gasca, J.; Greene, D.; Lee, D. S.; Muromachi, Y.; Newton, P. J.; Plotkin, S.; Sperling, D.; Wit, R.; Zhou, P. J. Transport and its infrastructure. In Climate Change 2007: Mitigation. Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Metz, B. et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, NY, 2007. Kandlikar, M. Air pollution at a hotspot location in Delhi: Detecting trends, seasonal cycles and oscillations. Atmos. Environ. 2007, 41 (28), 5934-5947. Kandlikar, M.; Reynolds, C.C.O.; Grieshop, A.P. A perspective paper on black carbon mitigation as a response to climate change. In Smart Solutions to Climate Change, Lomborg, B., Ed. Cambridge University Press: Cambridge, UK, (in press). Kim Oanh, N. T.; Upadhyay, N.; Zhuang, Y. H.; Hao, Z. P.; Murthy, D. V. S.; Lestari, P.; Villarin, J. T.; Chengchua, K.; Co, H. X.; Dung, N. T.; Lindgren, E. S. Particulate air pollution in six Asian cities: Spatial and temporal distributions, and associated sources. Atmos. Environ. 2006, 40 (18), 3367-3380. 37  Kirchstetter, T. W.; Harley, R. A.; Kreisberg, N. M.; Stolzenburg, M. R.; Hering, S. V. On-road measurement of fine particle and nitrogen oxide emissions from light- and heavy-duty motor vehicles. Atmos. Environ. 1999, 33 (18), 2955-2968. Kittelson, D. B. Engines and nanoparticles: A review. J. Aerosol Sci. 1998, 29 (5/6), 575-588. Lloyd, A. C.; Cackette, T. A. Diesel engines: Environmental impact and control. J. Air Waste Manag. Assoc. 2001, 51 (6), 809-847. Lough, G. C.; Christensen, C. G.; Schauer, J. J.; Tortorelli, J.; Mani, E.; Lawson, D. R.; Clark, N. N.; Gabele, P. A. Development of molecular marker source profiles for emissions from onroad gasoline and diesel vehicle fleets. J. Air Waste Manag. Assoc. 2007, 57 (10), 1190-1199. Lough, G. C.; Schauer, J. J. Sensitivity of source apportionment of urban particulate matter to uncertainty in motor vehicle emissions profiles. J. Air Waste Manag. Assoc. 2007, 57 (10), 12001213. Maricq, M. M.; Chase, R. E.; Xu, N.; Podsiadlik, D. H. The effects of the catalytic converter and fuel sulfur level on motor vehicle particulate matter emissions: Gasoline vehicles. Environ. Sci. Technol. 2002a, 36 (2), 276-282. Maricq, M. M.; Chase, R. E.; Xu, N.; Laing, P. M. The effects of the catalytic converter and fuel sulfur level on motor vehicle particulate matter emissions: Light duty diesel vehicles. Environ. Sci. Technol. 2002b, 36 (2), 283-289. Mazzi, E. A.; Dowlatabadi, H. Air quality impacts of climate mitigation: UK policy and passenger vehicle choice. Environ. Sci. Technol. 2007, 41 (2), 387-392. Mazzoleni, C.; Kuhns, H. D.; Moosmüller, H.; Witt, J.; Nussbaum, N. J.; Oliver Chang, M.; Parthasarathy, G.; Nathagoundenpalayam, S. K. K.; Nikolich, G.; Watson, J. G. A case study of real-world tailpipe emissions for school buses using a 20% biodiesel blend. Sci. Tot. Environ. 2007, 385 (1-3), 146-159. McCormick, R. L.; Graboski, M. S.; Alleman, T. L.; Alvarez, J. R.; Duleep, K. G. Quantifying the emission benefits of opacity testing and repair of heavy-duty diesel vehicles. Environ. Sci. Technol. 2003, 37 (3), 630-637. McTaggart-Cowan, G. P.; Reynolds, C. C. O.; Bushe, W. K. Natural gas fuelling for heavy-duty on-road use: current trends and future direction. Int. J. Environ. Stud. 2006, 63 (4), 421-440. MECA. Retrofitting Emission Controls for Diesel-Powered Vehicles. Manufacturers of Emission Controls Association: Washington, DC, 2009; www.meca.org/cs/root/resources/publications. Mehta, R. History, politics and technology of CNG-diesel switch in Delhi. Land Use, Transportation and the Environment; The Transport Asia Project: Pune, India, 2001; http://www.seas.harvard.edu/TransportAsia/Dec01Papers.htm.  38  Molina, M. J.; Molina, L. T. Megacities and atmospheric pollution. J. Air Waste Manag. Assoc. 2004, 54 (6), 644-680. MORTH. Motor Transport Statistics. Ministry of Road Transport and Highways: Delhi, India, 2007; http://www.morth.nic.in. Oliver, B. Greenhouse gas emissions and vehicle fuel efficiency standards for Canada. Pollution Probe: Toronto, ON, 2005. O'Neill, M. S.; Kinney, P. L.; Cohen, A. J. Environmental equity in air quality management: Local and international implications for human health and climate change. J. Toxicol. Env. Heal. A. 2008, 71 (9-10), 570-577. Pope, C. A., III. Air pollution and health - good news and bad. New Engl. J. Med. 2004, 351 (11), 1132-1134. Pope, C. A., III; Dockery, D. W. Health effects of fine particulate air pollution: Lines that connect. J. Air Waste Manag. Assoc. 2006, 56 (6), 709-742. Pope, C. A., III; Burnett, R. T.; Krewski, D.; Jerrett, M.; Shi, Y.; Calle, E. E.; Thun, M. J. Cardiovascular mortality and exposure to airborne fine particulate matter and cigarette smoke: Shape of the exposure-response relationship. Circulation. 2009, 120 (11), 941-948. Pucher, J.; Korattyswaropam, N.; Mittal, N.; Ittyerah, N. Urban transport crisis in India. Transp. Pol. 2005, 12 (3), 185-198. Rabl, A. Environmental benefits of natural gas for buses. Transp. Res. D – Tr. E. 2002, 7 (6), 391-405. Ramanathan, V.; Carmichael, G. Global and regional climate changes due to black carbon. Nature Geo. 2008, 1 (4), 221-227. Reynolds, C. C. O.; Kandlikar, M. Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860-5865. Rodgers, A.; Vaughan, P.; Prentice, T.; Edejer, T. T.; Evans, D.; Lowe, J. The World Health Report 2002: Reducing Risks, Promoting Healthy Life. World Health Organization: Geneva, Switzerland, 2002. Rogers, J. Assessment of the pollution under control program in India and recommendations for improvement. World Bank, South Asia Urban Air Quality Management Program: Washington, DC, 2002: http://siteresources.worldbank.org/PAKISTANEXTN/Resources/UrbanAir/MainReport.pdf. Rogers, K.; Seager, T. P. Environmental Decision-Making Using Life Cycle Impact Assessment and Stochastic Multiattribute Decision Analysis: A Case Study on Alternative Transportation Fuels. Environ. Sci. Technol. 2009, 43 (6), 1718-1723. 39  Ropkins, K.; Beebe, J.; Li, H.; Daham, B.; Tate, J.; Bell, M.; Andrews, G. Real-World Vehicle Exhaust Emissions Monitoring: Review and Critical Discussion. Crit. Rev. Environ. Sci. Technol. 2009, 39 (2), 79-152. Sakai, T.; Nakajima, T.; Yamazaki. H. O-PM/Emitted matters caused by two-stroke engine oil and its reduction. SAE Tech. Pap. 1999, 1999-01-3260. Samulski, M.; Jackson, C. Effects of steady-state and transient operation on exhaust emissions from nonroad and highway diesel engines. SAE Tech. Pap. 1998, 982044. Searchinger, T.; Heimlich, R.; Houghton, R.; Dong, F.; Elobeid, A.; Fabiosa, J.; Tokgoz, S.; Hayes, D.; Yu, T. Use of US croplands for biofuels increases greenhouse gases through emissions from land-use change. Science. 2008, 319 (5867), 1238-1240. Shine, K. P.; Berntsen, T. K.; Fuglestvedt, J. S.; Skeie, R. B.; Stuber, N. Comparing the climate effect of emissions of short- and long-lived climate agents. Philos. Trans. R. Soc. London, Ser. A. 2007, 365 (1856), 1903-1914. SIAM. Overview of performance of the Indian Automobile Industry in 2007-2008. Society of Indian Automobile Manufacturers: New Delhi, India; www.siamindia.com/scripts/industrystatistics.aspx Singh, A.; Gangopadhyay, S.; Nanda, P. K.; Bhattacharya, S.; Sharma, C.; Bhan, C. Trends of greenhouse gas emissions from the road transport sector in India. Sci. Tot. Environ. 2008, 390 (1), 124-131. Smith, K. R.; Haigler, E. Co-benefits of climate mitigation and health protection in energy systems: Scoping methods. Ann. Rev. Pub. Health, 2008, 29, 11-25. Smith, K. R.; Jerrett, M.; Anderson, H. R.; Burnett, R. T.; Stone, V.; Derwent, R.; Atkinson, R. W.; Cohen, A.; Shonkoff, S. B.; Krewski, D.; Pope, C. A., III; Thun, M. J.; Thurston, G. Health and Climate Change 5 – Public health benefits of strategies to reduce greenhouse-gas emissions: health implications of short-lived greenhouse pollutants. Lancet. 2009, 374 (9707), 2091-2103. Smith, M. Advances in understanding benzene health effects and susceptibility. Annu. Rev. Public. Health. 2010, 31, 133-148. Srivastava, A.; Jain, V. Source apportionment of suspended particulate matters in a clean area of Delhi: A note. Transp. Res. D – Tr. E. 2008, 13 (1), 59-63. Stasko, T. H.; Gao; H. O. Reducing transit fleet emissions through vehicle retrofits, replacements, and usage changes over multiple time periods. Transp. Res. D. – Tr. E. 2010, 15 (5), 254-262. St. Denis, M.; Lindner, J. Review of light-duty diesel and heavy-duty diesel gasoline inspection programs. J. Air Waste Manag. Assoc. 2005, 55 (12), 1876-1884.  40  Subramanian, R.; Winijkul, E.; Bond, T. C.; Thiansathit, W.; Kim Oanh, N. T.; Paw-Armart, I.; Duleep, K. Climate-relevant properties of diesel particulate emissions: Results from a piggyback study in Bangkok, Thailand. Environ. Sci. Technol. 2009, 43 (11), 4213–4218. Swart, R.; Amann, M.; Raes, F.; Tuinstra, W. A good climate for clean air: Linkages between climate change and air pollution. Climatic Change. 2004, 66 (3), 263-269. TERI. TERI Energy Data Directory and Yearbook 2004/05. The Energy and Resources Institute: New Delhi, India, 2006. Thornhill, D. A.; Williams, A. E.; Onasch, T. B.; Wood, E.; Herndon, S. C.; Kolb, C. E.; Knighton, W. B.; Zavala, M.; Molina, L. T.; Marr, L. C. Application of positive matrix factorization to on-road measurements for source apportionment of diesel- and gasoline-powered vehicle emissions in Mexico City. Atmos. Chem. Phys. 2010, 10, 3629-3644. Twigg, M. V. Progress and future challenges in controlling automotive exhaust gas emissions. Appl. Catal. B – Environ. 2007, 70 (1-4), 2-15. UNEP. Asia-Pacific Diesel Sulphur Matrix. UN Environment Program Partnership for Clean Fuels and Vehicles, UN Environment Program (UNEP); www.unep.org/pcfv/regions/Asiapacific.asp. USEPA. A Comprehensive Analysis of Biodiesel Impacts on Exhaust Emissions. Report No. EPA420-P-02-001; US Environmental Protection Agency: Washington, DC, 2002. USEPA. The Particle Pollution Report: Current Understanding of Air Quality and Emissions through 2003. Report No. EPA454-R-04-002; US Environmental Protection Agency: Washington, DC, 2004. Van Houtte, J.; Niemeier, D. A critical review of the effectiveness of I/M programs for monitoring PM emissions from heavy duty vehicles. Environ. Sci. Technol. 2008, 42 (21), 78567865. Volckens, J.; Braddock, J.; Snow, R. F.; Crews, W. Emissions profile from new and in-use handheld, 2-stroke engines. Atmos. Environ. 2007, 41 (3), 640-649. Volckens, J.; Olson, D. A.; Hays, M. D. Carbonaceous species emitted from handheld two-stroke engines. Atmos. Environ. 2008, 42 (6), 1239-1248. Warnatz, J.; Maas, U.; Dibble, R.W. Combustion: Physical and Chemical Fundamentals, Modeling and Simulation, Experiments, Pollutant Formation, 2nd edition, Springer: Berlin, 1999. West, J. J.; Fiore, A. M.; Horowitz, L. W.; Mauzerall, D. L. Global health benefits of mitigating ozone pollution with methane emission controls. PNAS. 2006, 103 (11), 3988-3993.  41  Wong, C.; Vichit-Vadakan, N.; Kan, H.; Qian, Z. Public health and air pollution in Asia (PAPA): A multicity study of short-term effects of air pollution on mortality. Environ. Health Persp. 2008, 116 (9), 1195-202. World Bank. World Bank Indicators: Motor Vehicles (per 1,000 people). World Bank: Washington, DC; http://data.worldbank.org. Yanowitz, J.; McCormick, R. L.; Graboski, M. S. In-use emissions from heavy-duty diesel vehicles. Environ. Sci. Technol. 2000, 34 (5), 729-740. Yuan, D.; Zhou, W.; Ye, S. Effect of leaded and unleaded gasoline on the mutagenicity of vehicle exhaust particulate matter. J. Environ. Pathol. Toxicol. Oncol. 2000, 19 (1-2), 41-8. Zhao, J. Whither the car? China's automobile industry and cleaner vehicle technologies’, Devel. Change. 2006, 37 (1), 121-144.  42  2. Chapter 2: Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in Delhi5  2.1. Introduction  Compressed natural gas (CNG) fueling for vehicles is seen as a means of reducing environmental and human health costs of transportation, since internal combustion engines running on CNG produce inherently less pollutant emissions than comparable liquid-fuel engines. This is especially attractive to developing nations, where advanced liquid-fuel vehicles with low-emitting engines and tailpipe pollution controls may not be affordable. Various jurisdictions, for example, New Delhi, Mumbai, Mexico City and Rio de Janeiro, have already converted vehicle fleets to natural gas fueling (Balassiano and White 1997; Schifter et al. 2000). When conventional vehicles, especially heavy-duty trucks and buses fueled by diesel, are converted to natural gas, mass emissions of particulate matter (PM) can be reduced by one to two orders of magnitude (Rabl 2002). This is certainly important from a public health perspective, because PM emissions from diesel engines are carcinogenic and cause cardiopulmonary health effects (Pope and Dockery 2006) and the health impacts of PM concentrations in Delhi are well known (Cropper et al. 1997). Although an analysis of the health effects of the fuel-switch is outside the scope of the present study, air pollution and climate change are inextricably connected because the combustion of fossil fuels releases greenhouse gases, aerosols and other criteria air contaminants. Policy analyses that take into account potential climate/air pollution synergies have been called for, and are clearly needed (Swart et al. 2004). The present study takes this approach, by quantifying the change in climate-forcing emissions that result from a fuel-switching policy that was designed with air quality in mind.  5  A version of this chapter has been published: Reynolds, C. C. O.; Kandlikar, M. Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860-5865.  43  One of the most significant fuel-switching examples has been in Delhi, where the entire public transportation fleet – buses, taxis and auto-rickshaws (three-wheeled motorcycle taxis) – were converted to run entirely on natural gas (Mehta 2001). The switch was mandated by a Supreme Court of India directive (as a response to a public petition), which required that the Delhi Government tackle the problem of air pollution from public transport vehicles. Initially scheduled to be completed in 2001, technical difficulties with retrofitting so many vehicles (and the limited number of NG fueling stations) caused the process to be completed only in 2003. Although there has been wide acclaim for this move, the jury is still out regarding the magnitude of air quality improvements that resulted (Goyal and Sidhartha 2003; Kandlikar 2007; Kathuria 2004; Ravindra et al. 2006; Narain and Krupnick 2007). In one of the most recent analyses, spectral analysis methods were used to examine changes in pollutant concentrations since 2000 at a ‘hotspot’ for traffic pollution in New Delhi (Kandlikar 2007). Ambient levels of carbon monoxide (CO) and sulfur oxides (SOx) were observed to have been reduced, coincident with the switch to CNG fuel, but PM10 (PM with an aerodynamic diameter of less than 10 microns) concentrations remained essentially unchanged over the whole period, and nitrogen oxide concentrations rose until 2004, followed by a decline thereafter. There is no doubt that public transportation emissions changed following the fuel switch. However, the signal to noise ratio (as evidenced in the air concentration data) appears to be too low for the effects of the policy to be discernible from other sources such as industry and power generation, for all pollutants with the exception of carbon monoxide. One aspect of this fuel change in New Delhi that has thus far been left unexplored is the climate forcing implications of switching to CNG. This work aims to fill that gap.  2.2. Methods  2.2.1. Fuel efficiency and emissions factors As of April 2005, there were almost 90,000 public transportation vehicles in the New Delhi metropolitan area operating on CNG fuel (TERI 2006). Annual average vehicle activity data (km per year) for each type of vehicle are not well characterized in India. Only one vehicle study was found to have measured the daily distance traveled for a randomly selected statistical sample in India. The study was conducted in the Indian city of Pune in 2003 (ISSRC 2004) and yielded 44  results similar to estimated activity data for different transportation categories in Indian cities (Kandlikar and Ramachandran 2000), and for buses in Mumbai (Takeuchi et al. 2007). We assumed that public transportation vehicle activity in Delhi was similar to that in Pune, while also recognizing that the difference between CNG and gasoline/diesel emissions on a per-vehicle basis is not sensitive to activity levels. The total activity for each category was then calculated (see Table A.1 in Appendix A). Although an average auto-rickshaw travels the lowest distance annually, this category makes up the greatest number of CNG vehicles by a factor of five, and hence represents the greatest total activity. The CNG vehicles were assumed to be all direct retrofits (rebuilt engines and fueling systems) from the original diesel and gasoline vehicles, rather than new vehicles with purpose-built CNG engines. In New Delhi, resources were not available to purchase large numbers of new CNG vehicles prior to 2005, and even in the future, ‘new’ CNG vehicles are likely to be technologically equivalent to retrofitted vehicles due to the prohibitive expense of state-of-theart CNG-engine technology. Consequently, because vehicle weight, engine size, transmission, aerodynamics, and general mechanical condition are likely to be unchanged, the CNG vehicles’ fuel consumption can be assumed to be proportional to that of pre-conversion vehicles. However, some fuel efficiency losses must be attributed to the conversion, in particular for the heavy-duty CNG engines converted from diesel engines. A throttle must be added in the air intake to control the engine power, resulting in a significant efficiency loss (diesel engines do not need a throttle; simply changing the amount of injected fuel results in more or less engine power output). Other efficiency losses arise from sub-optimal design related to the following: the sparkignition system that must be installed in retrofitted diesel bus engines; inappropriate combustion chamber design for its new fuel; the compression ratio being too low in the gasoline engines (natural gas has a much higher octane number than gasoline); low CNG pressures; the extra mass of CNG tanks; and problems with the fuel carburetor/injectors. There is a lack of specific technical information published about the retrofitted fleet in New Delhi. Therefore, we assumed a 25% fuel efficiency penalty for the bus conversions, which is typical of diesel-CNG conversions described in the literature (Kojima 2001). For gasoline-CNG conversion, we assume only 5% fuel efficiency penalty since these vehicles already have sparkignition and hence do not suffer additional throttling losses (Bhangale and Ghosh 1995). CNG 45  has higher hydrogen to carbon ratio than liquid hydrocarbon fuels, so it produces less CO2 per unit energy released during combustion, and this partially offsets the loss in fuel efficiency from conversion. Average fuel consumption values (expressed as kilograms per 100km so that a comparison can be made between liquid and gaseous fuels) and CO2 emissions factors are given in the Table A.2 (Appendix A). CO2 emissions factors for each vehicle/fuel type have been derived using the assumption that all fuel is completely burned. In reality a small fraction of fuel carbon is emitted in the form of carbon monoxide (CO) and volatile organic compounds. However, CO is ultimately oxidized to CO2 in the atmosphere, so this source of uncertainty has no impact on the CO2 emissions factors. We capture the impact of volatile organic compounds through the effect of organic aerosol PM on atmospheric radiative forcing. Net CO2 emission factors increase by about 9% after diesel-CNG conversion, but are approximately 8% lower after gasoline-CNG conversion. Non-CO2 climate-forcing emissions addressed in this study fall into two categories: (a) methane emissions, from both ‘evaporative leakage’ of natural gas from the CNG vehicles only, and as a component of the exhaust from both liquid-fuel and CNG vehicles; and (b) aerosol emissions (BC, OC and sulfate) from both liquid-fuel and CNG vehicles. Nitrous oxide (N2O), another potent greenhouse gas, is not included here because net mass emissions of this species are not appreciably different for diesel vs. CNG engines (Lipman and Delucchi 2002). Refrigerants in vehicle air-conditioning systems (such as HCFCs) are strongly climate-forcing species, and are problematic if leaked to the atmosphere. However we assume that there is no net change in refrigerant leakage attributable to the fuel-switching policy. Nitrogen oxide (NOX) emissions can also have an indirect climate forcing impact, via mechanisms that form nitrates, shorten the atmospheric lifetime of CH4 (both of which cause negative forcing), and the formation of ozone (positive forcing) (West et al. 2007). It is unclear therefore whether increased ambient NOX concentration will lead to increases or decreases in radiative forcing. Further, increases in NOX directly attributable to the fuel-switch are small (Kandlikar 2007) to negligible (Narain and Krupnick 2007). Consequently, we assume that NOX changes from fuel switching have a negligible climate impact. Table 2.1 summarizes the representative emissions factors for climate-forcing emissions of the three categories of public transportation vehicles, buses, taxis and auto-rickshaws, before and 46  after the switch to CNG fueling. In particular, the published CH4 emissions factors for retrofitted CNG vehicles are highly uncertain. We have chosen to use medium emissions factors for both CH4 and PM from the range found in the literature, and then explore the implications of varying these emissions across the range of uncertainty. Details on the derivation of emissions factors for CH4 and aerosols are provided in the Supporting Information (Appendix A, Section A.2.1).  Table 2.1. Summary of climate-forcing emissions factors for liquid-fuel and CNG public transportation vehicles: gaseous and aerosol species.  Gaseous emissions: Carbon Dioxide Methane (exhaust) Methane (leakage) Aerosols: Black Carbon Organic Carbon SO2  Liquid-Fuel Emissions Factors (g/km) AutoBuses Cars rickshaws 1063 157 67 0.06 0.14 0.08 0 0 0  CNG Emissions Factors (g/km) AutoBuses Cars rickshaws 1160 144 62 6.50 2.28 1.30 1.99 0.25 0.11  1.52 0.48 0.233  0.002 0.005 0  0.16 0.17 0.015  0.01 0.19 0.006  0.001 0.003 0  0.008 0.024 0  2.2.2. Global warming/cooling metrics for climate forcing aerosols Different types of particles in the atmosphere reflect or absorb radiation depending on their optical properties. Light colored sulfate and organic carbon aerosols reflect solar radiation, which has a cooling effect (Forster et al. 2007). They are also understood to cause indirect cooling, through increased cloud albedo. In contrast, black carbon (BC) particulate matter absorbs light, and consequently warms the atmosphere. In addition to direct and indirect atmospheric radiative forcing effects, black carbon deposited on snow and ice reduces the albedo of the frozen surface, which has been shown to accelerate melting rates (Hansen and Nazarenko 2004). Recent research has demonstrated that there are regional and global climate impacts of atmospheric black carbon (BC) (Menon et al. 2002; Ramanathan et al. 2007), and it has been proposed that control of BC emissions could be an economical means of reducing anthropogenic climate impacts, especially in rapidly industrializing countries (Bond and Sun 2005; Jacobson 2001; Shine et al. 2007).  47  Tropospheric aerosols are relatively short-lived, and remain in the atmosphere for weeks rather than years. Despite significant uncertainty regarding the climatic effects of aerosols, a number of studies have presented calculations of the global warming potential (GWP) for atmospheric BC. Hansen and colleagues (Hansen et al. 2007) have calculated a GWP for fossil-fuel derived BC of 500, which includes both positive forcing from soot particles, as well as negative forcing (direct and indirect) from co-emitted organic carbon (OC). This value compares well with other published values: Bond & Sun calculate a GWP for BC of 680 for the same time horizon, not including the effect of co-emitted organic carbon (Bond and Sun 2005). There are some lower GWP estimates, for example 90-190 for BC plus OC (Jacobson 2002; 2005), and 120-230 for BC only (Berntsen et al. 2006). However, these studies tend to underestimate forcing effects (Bond and Sun 2005; Bond 2007). Although the cooling effect of sulfates and organic carbon has long been recognized (Houghton et al. 1990), a GWP-like metric to represent their climatic impact has not yet been developed. In order to evaluate the climatic impacts of aerosols with respect to carbon dioxide, we need a value for their global warming or cooling potential. For this study, we have estimated the GWP for BC using radiative forcing and atmospheric lifetime information presented in the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (Forster et al. 2007). This approach allows us to also estimate a cooling metric for SO2 and OC, which we refer to as Global Cooling Potential (GCP). IPCC defines GWP for component i as: TH  TH  ! RFi (t )dt  GWPi # TH0  !a  " [Ci (t )]dt  = TH0  ! RF (t )dt ! a r  0  i  r  (Equation 2.1.) " [C r (t )]dt  0  where TH is the time horizon (typically set to 100 years), RFi is the global mean radiative forcing (RF) of component i, ai is the RF per unit mass increase in atmospheric abundance of component i (radiative efficiency), Ci(t) is the time-dependent abundance of a single unit of i, emitted to the atmosphere at t = 0. The corresponding quantities for the reference gas (r) are in the  48  denominator. GWPs are normally calculated with CO2 as the reference gas. Equation 2.1 is equivalent to:  GWPi RFi /Si = GWPCO2 RFCO2 /SCO2  !  (Equation 2.2.)  where RFi is the integrated RF contribution (over 100 years) of a single emission pulse of magnitude Si released at t=0. To calculate GWP/GCP for aerosols, the best estimates for RFi (including indirect effects) are given for the year 2000 in the IPCC’s Fourth Assessment Report (Forster et al. 2007), hence Si is the source strength (Tg year-1) of species i for that year. RFBC ,  RFOC and RFS are estimated to be 0.26 (0.11 to 0.41) Wm-2, -0.10 (-0.02 to -0.25) Wm-2, and 0.91 (-0.42 to -1.95) Wm-2 respectively for year 2000 global emissions (uncertainty bounds in parentheses after each value). Average source strengths for BC, OC and SO2 were obtained from the Aerocom experiment used in the IPCC calculations (Dentener et al. 2006), and were approximately 6.32, 32.5 and 100.7 Tg year-1 respectively for year 2000 emissions. Carbon dioxide has a RF of 2.40±0.4 Wm-2, and its source strength in 2000 was 26,400 Tg year-1. Using equation 2.2, we calculate the mean GWP for BC to be 455 (193 to 716), which compares well with other estimates. The mean GCPs for SO2 and OC are calculated as being -100 and -35 respectively. The 95% Confidence intervals for the aerosols’ GWP/GCPs have been calculated from uncertainties in RFi , and are summarized in the Supporting Information (Appendix A, Table A.4). The implications of these uncertainties for our results are discussed in section 2.3.3. Finally, we note that while there are undoubtedly challenges with using metrics such as GWP/GCP to include aerosols in global climate agreements, a detailed discussion of this topic is outside the scope of this paper. Interested readers should consult Bond (2007) for an assessment of some of the barriers to more comprehensive climate agreements, and arguments to overcome them.  49  2.3. Results  2.3.1. Change in CO2-equivalent emissions Climate-forcing emissions can all be converted to a common metric, namely units of carbon dioxide equivalent emissions (CO2-eq). In this way, a comprehensive assessment of the climatic impacts of the CNG fuel switch can be made, that includes all climate-forcing species. CO2-eq emissions are calculated using the following formula: CO2 (e) = " AV # NV # EFV ,i # Pi  (Equation 2.3.)  V ,i  !  where NV and AV are the numbers and average activity (kilometres per year per vehicle) of vehicle type V (buses, cars or auto-rickshaws), EFV,i is the emission factor for emissions species i from an average vehicle of type V, and Pi is the global warming or cooling potential of that species with respect to the reference species, carbon dioxide. CO2-eq emissions are reported by vehicle and by species, for before and after the switch to CNG fueling, in Appendix A (Table A.5). If only carbon dioxide and methane are considered, overall we find that there is approximately a 30% increase in conventional greenhouse gases (GHGs) attributable to the switch to CNG fueling in New Delhi. Although we argue that aerosols should be included in the assessment of climate impacts, we start from this point because aerosols are not currently recognized under the Kyoto Protocol. All vehicle categories show an increase in conventional GHGs, in part due to the increase in fuel consumption, but primarily due to exhaust and leakage emissions of methane. However, the inclusion of aerosol emissions has a very important impact on radiative forcing. This is illustrated in Figure 2.1, which graphically summarizes the results of this study.  50  ! CO2-eq (103 tonnes) ! CO2-eq (103 tonnes)  ! CO2-eq (103 tonnes)  A.  B.  C.  D.  Figure 2.1. Emission inventories, demonstrate the change in climate-forcing emissions attributable to the switch from diesel- and gasoline-fueled vehicles to CNG vehicles. Units are 103 tons of CO2 equivalent emissions (CO2-eq). A. All climate forcing emissions, including black carbon, particulate organic carbon and sulfur dioxide (precursor to sulfate particulate) aerosol species, are included. B-D. Change in climate-forcing emissions (!CO2-eq) due to fuelswitching buses, cars and auto-rickshaws respectively. Note that the vertical scale on panel B is twice that of panels C and D.  Figure 2.1.A represents the results aggregated by vehicle type, and Figures 2.1.B-D illustrate the breakdown of climate-forcing emissions in more detail for buses, cars and auto-rickshaws. After conversion to CNG-fueling, buses (Figure 2.1.B) emit more direct CO2 emissions, and more CO2-eq due to an increase in CH4 from almost zero, but the CO2-eq from BC is very significantly reduced. OC and SO2 emissions have very little effect on the results. Both cars and auto-rickshaws (Figures 2.1.B and C] exhibit a reduction in direct CO2 emissions, due to similar fuel consumption of retrofitted spark-ignition engines and the lower carbon content of methane. The climate impact of post-conversion methane emissions is significant for both vehicle types, but more so for auto-rickshaws because they are likely to be fitted with less high-technology 51  engines. The CO2-eq reduction attributable to reducing BC emissions from cars is significant and about the same order of magnitude (though with opposite sign) to the impact of methane emissions. Auto-rickshaws’ CO2-eq emissions are affected by the reduction in BC and OC. The reduction in reflective OC particulate matter reduces its cooling effect, and so actually increases the net CO2-eq emissions. However, OC is strongly suspected to have important human health impacts (McDonald et al. 2004), so its reduction is an important outcome of CNG policy, climate impacts notwithstanding. As with diesel buses, SO2 has little impact on net CO2-eq emissions from gasoline cars or auto-rickshaws. When aerosol emissions are included, the switch to CNG fueling results in a climate benefit, largely because of the dramatic reduction of black carbon emissions from the diesel bus engines. In total there is about a 10% reduction of net CO2-eq emissions, and if buses are considered separately, net CO2-eq emissions are reduced by about 20%. In a similar manner, if cars and auto-rickshaws are considered as independent subgroups, fuel switching results in a net reduction in CO2-eq of approximately 10% for cars, and a net increase of about 50% for autorickshaws. In the case of auto-rickshaws, the net CO2-eq increase is primarily due to the significantly increased exhaust emissions of unburned methane that occurs when vehicles are converted to run on natural gas. Auto-rickshaws are a special case, since there were a high proportion of 2-stroke gasoline engines prior to conversion, most of which would have been scrapped and replaced with new vehicles. The majority of CNG auto-rickshaws are assumed to operate on a 4-stroke cycle, producing inherently lower mass emissions of particulate matter. Overall, black carbon emissions from diesel buses dominate the aerosols’ contribution, and methane emissions from converted vehicles are also very important. The results are very sensitive to uncertainty in these emissions factors. We analyze this further in the following section.  2.3.2. Uncertainty in emissions factors The results presented above are based on the best data about emissions factors currently available in the literature. The use of average emissions factors (and, indeed, average annual vehicle activity) is a significant simplification of reality, given the enormous variation in vehicle types and conditions. Emissions factors are highly sensitive not only to the engine technology, but also to influences as diverse as driver behaviour, vehicle loading, fuel quality, local 52  topography, climatic conditions, and traffic conditions. Ideally, emissions factors would be based on direct measurement of the exhaust from representative vehicles being driven over a realistic drive-cycle, but this information is not generally available for countries outside the Organization for Economic Co-operation and Development (OECD), and even within the OECD, few emissions models use this approach. Consequently, uncertainty in the emissions factors could have considerable impact on our results, namely the change in net CO2-eq emissions resulting from the CNG switch. This is particularly true for methane emissions factors from the retrofitted buses and particulate matter emissions from pre-conversion diesel and gasoline vehicles because (i) there is a significant change in the emissions rate of these species, and (ii) the global warming/cooling potentials for these species are high. We have investigated the sensitivity of our model results (the change in net CO2-eq emissions) to variation in the CH4 and PM emissions factors inputs, as shown in Figure 2.2. The range of uncertainty in these emissions factors was estimated to be on the order of factor of three. For the diesel buses, ‘low’, ‘medium’ and ‘high’ PM emissions factors corresponded to 1.0, 2.0, and 3.0 g/km of total PM respectively (76% of PM from buses was assumed to be BC and the remainder OC). The ‘low’, ‘medium’ and ‘high’ CH4 emissions factors for the retrofitted CNG buses corresponded to 3.0, 6.5, and 10.0 g/km of CH4 respectively. The range of emissions factors tested for cars and auto-rickshaws were scaled proportionally to the range of values for buses. The results of this analysis indicates that there is more likely to be a net reduction in CO2-eq emissions. The shaded area in Figure 2.2 is 85% of the total area, indicating that a net reduction could be 5 times more likely than a net increase (unshaded area) once uncertainties in aerosol emissions are accounted for.6 This supports our conclusion that there may be significant climate benefits of switching to CNG fueling when the aerosol forcing effect is included in the calculations. Furthermore, if PM emissions from the old diesel engines are in fact ‘medium’ or ‘high’, then there is a net climate benefit of the fuel-switch, no matter what the CH4 emissions factors are.  6  Probabilistically, emission factors at the extremes of the considered ranges are less likely than central estimates – therefore this sensitivity analysis is a simplification of a true uncertainty analysis.  53  -eq  Figure 2.2. Sensitivity of model results to emission factors: effect on change in net CO2-eq emissions when CH4 and PM emissions factors are varied around the ‘medium’ emissions factors used for the detailed analysis described in section 2.3.1 (indicated here by the black point in the centre of the figure). The x and y axes refer to CH4 and PM emissions factors respectively. The contours on the graph indicate the percentage change in CO2-eq after the CNG switch, for the range of CH4 and PM emissions factors tested (‘low’, ‘medium’ and ‘high’ values), and the shading shows the area of net climate benefit (CO2-eq reduction).  2.3.3. Uncertainty in GWP/GCP There is an ongoing debate about the influence of various parametric and other uncertainties on GWP values of the major non-CO2 greenhouse gases (Shine et al. 2007; Kandlikar 1995; Hayhoe et al. 2000). However, under the Kyoto protocol CO2 equivalence is established using a single representative value for the GWP (100 year horizon) of the non-CO2 greenhouse gases, which is the approach we take in this paper. The impact of uncertainties in GWP/GCP of aerosols was investigated, and it was found that its effect on our results depends directly on the net annual emissions of each aerosol species. For cooling aerosols (OC and SO2) the net emissions from public transportation vehicles are small when compared to emissions of CO2, BC and CH4 emissions. Consequently, even large uncertainties in the GCP of OC and SO2 (resulting primarily from indirect forcing effects of the cloud-albedo feedback) have little impact on our conclusion. Uncertainties in the GWP of BC are more important. To examine their effect we replicated the sensitivity analysis (described in section 2.3.2) using the lower (GWP = 193) and upper (GWP = 716) 95% confidence bounds. For the lower bound GWP value for BC, net 54  reductions are possible only when emission factors for PM are ‘medium’ or greater and when CH4 emissions factor is lower than ‘medium’. For the upper bound of GWP = 716, net reductions of CO2-eq emissions occur almost independent of the either set of emission factors (see Figure A.1 in Appendix A).  2.4. Discussion  Our analysis demonstrates that the Indian Supreme Court-mandated policy, to switch New Delhi’s public transportation to CNG fueling in 2002, resulted in a very substantial increase in CO2 and CH4 emissions. However, in the light of recent research about the climatic impacts of atmospheric aerosols, we argue that it is essential to consider particulate matter emissions in the assessment of this policy. We find that the fuel-switching policy resulted in a dramatic reduction in BC emissions from buses. Therefore, when we include aerosols, the climate impact results change from ‘strong positive forcing’ to ‘neutral/strong negative forcing’. Sulfates and organic carbon from diesels and gasoline vehicles have a global cooling effect, although the magnitude of their impact relative to BC is small. Our findings confirm the assertion by Bond and Sun (2005) and others that addressing BC emissions from public transport is likely to be a promising way to reduce climate interference. Methane emissions factors from the retrofitted vehicles figure prominently in net climate forcing calculations, so emissions of methane from CNG vehicles may also provide a near-term opportunity for reducing climate interference. The potential benefit of methane reduction can be understood by referring back to Figure 2.2: a reduction in methane emissions factors is equivalent to reducing uncertainty and finding that they are ‘low’ rather than ‘medium’. The net climate benefit would increase to 20% from 10%. Tailpipe methane emissions are not regulated in many jurisdictions, so there is little incentive for engine technology providers to target reductions in the amount of methane emitted in the exhaust. Consequently, there may be an opportunity to further reduce the climate impacts of fuel switching by stipulating reasonable emissions levels for methane from CNG vehicles. Inspection and maintenance programs are one way of ensuring that all engines are tuned for low emissions, but such programs can be expensive and administratively challenging. Another possibility would be to ensure all retrofitted vehicles are equipped with three-way catalytic converters, which reduce emissions of NOX and 55  CO as well as CH4, but retrofitted catalysts have durability issues on older engines. Replacing the current standard of retrofitted engines with entirely new CNG engines (or improved retrofits) may represent an opportunity for emissions reduction, and may be fundable under the UN Framework Convention on Climate Change (UNFCCC) Clean Development Mechanism (CDM). Such reductions would meet the additionality criteria and would be an easily verifiable source of carbon credits, although BC reductions would not be admissible under current CDM rules (UNFCCC 2007). In Table 2.2 we show the value of carbon credits for two levels of improvement in methane emissions from CNG vehicles, assuming the base case emissions factors are those used in this study: the ‘100% reduction’ case refers to a reduction of methane emissions in the exhaust to levels that are negligible from a climate perspective, which is possible using state of the art CNG engine technology but may not be viable in rapidly industrializing countries due to high cost. For the scenario where a 45% reduction below original methane emissions factors is realized (achievable using readily-available technology), the net present value of reduced CO2-eq emissions over the vehicles’ lifetimes is approximately US$1,463 per bus, US$313 per car and US$121 per auto-rickshaw, assuming a value per tonne of CO2-eq that starts at $20 and grows annually by 5%, and a discount rate of 10%. It is clear there are substantial economic incentives to reduce methane tailpipe emissions by optimizing CNG engine design to meet low methane emission criteria. The estimated value of these carbon credits is sufficient to justify further investigation into the cost of CNG-engine upgrades and applicability to the CDM. Table 2.2. Net present value of carbon credits due to replacing retrofitted CNG engines with new (or improved) CNG engines. Two mitigation options are shown, corresponding to 45% and 100% reduction in CH4 emissions factors, respectively. Buses Cars Auto-rickshaws Change in CH4 Emissions Factors 45% reduction $1460 $310 $120 100% reduction (i.e. negligible CH4) $2720 $580 $230 Emissions reduction over vehicle lifetime (lifetime is assumed to be 20, 15, and 12 years for buses, cars, and autorickshaws respectively); 1 tonne CO2-eq is valued at US $20 (current dollars), and this value is assumed to increase by 5% per year (to $51 in year 20); discount rate is10%.  There are some lessons that may be drawn from the Delhi case for similar fuel-switching projects. First, buses (and other heavy-duty diesel vehicles) are more important than the other vehicle types to convert; this is because they emit significant BC compared to gasoline vehicles, and BC has adverse impacts from both health and climate perspectives. Second, since 56  eliminating OC from 2-stroke engines may have substantial health impacts, switching to CNG in 2-stroke auto-rickshaws may also be highly beneficial. Finally, exhaust methane emissions are very important contributors to climate forcing from CNG vehicle operation, so regulation of tailpipe methane emissions from new or retrofitted CNG vehicles would go a long way towards reducing their climate impact. More generally, there is an emerging literature on the tradeoffs between local and global benefits from fuel switching (Swart et al. 2004; Mazzi and Dowlatabadi 2007; Dowlatabadi 2007). Here we add to that literature by demonstrating that the climate impacts of policies put in place for reasons other than the climate need to be characterized, especially in rapidly industrializing countries where climate mitigation is not currently being implemented. The climate benefits of non-climate policies can be substantial, especially when aerosol emissions are included.  57  2.5. References Balassiano, R.; White, P. Experience of compressed natural gas bus operations in Rio de Janeiro, Brazil. Transport. Res. D-Transp. Environ. 1997, 2 (2), 147-155. Berntsen, T.; Fuglestvedt, J.; Myhre, G.; Stordal, F.; Berglen, T. Abatement of greenhouse gases: Does location matter? Climatic Change. 2006, 74 (4), 377-411. Bhangale, U. D.; Ghosh, B. ARAI experiences on conversion of petrol (gasoline) engine vehicles to CNG operation. SAE Tech. Pap. 1995, 950403. Bond, T. C. Can warming particles enter global climate discussions? Environ. Res. Lett. 2007, 2 (045030). Bond, T. C.; Sun, H. Can reducing black carbon emissions counteract global warming? Environ. Sci. Technol. 2005, 39 (16), 5921-5926. Cropper, M. L.; Simon, N. B.; Alberini, A.; Arora, S.; Sharma, P. K. The health benefits of air pollution control in Delhi. Amer. J. Agr. Econ. 1997, 79 (5), 1625-1629. Dentener, F., et al. Emissions of primary aerosol and precursor gases in the years 2000 and 1750, prescribed data-sets for AeroCom. Atmos. Chem. Phys. 2006, 6 (12), 4321-4344. Dowlatabadi, H. On integration of policies for climate and global change. Mitig. Adapt. Strateg. Glob. Change. 2007, 12 (5), 651-663. Forster, P., Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.; Haywood, J.; Lean, J.; Lowe, D.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Dorland, R. V. Changes in atmospheric constituents and in radiative forcing, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon, S. et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, 2007. Goyal, P.; Sidhartha. Present scenario of air quality in Delhi: a case study of CNG implementation. Atmos. Environ. 2003, 37 (38), 5423-5431. Hansen, J.; Nazarenko, L. Soot climate forcing via snow and ice albedos. Proc. Nat. Acad. Sci. 2004, 101 (2), 423-428. Hansen, J.; Sato, M.; Kharecha, P.; Russell, G.; Lea, D. W.; Siddall, M. Climate change and trace gases. Philos. Trans. R. Soc. London, Ser. A. 2007, 365 (1856), 1925-1954. Hayhoe, K.; Jain, A.; Keshgi, H.; Wuebbles, D. Contribution of CH4 to multi-gas reduction targets: The impact of atmospheric chemistry on multi-gas GWPs, in Non-CO2 Greenhouse Gases: Scientific Understanding, Control and Implementation (Proceedings of the Second International Symposium), J. van Ham, Editor. Kluwer Academic Publishers: Noordwijkerhout, Netherlands. 2000, p. 425–432. 58  Houghton, J. T.; Jenkins, G. J.; Ephraums, J. J., Eds. Climate Change 1990: Scientific Assessment of Climate change. Contribution of Working Group I to the First Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press: Cambridge, UK and New York, 1990. ISSRC, Pune vehicle activity study (March 9-March 22 2003). International Sustainable Systems Research Centre: La Habra, CA, 2004: http://issrc.org/. Jacobson, M. Z. Strong radiative heating due to the mixing state of black carbon in atmospheric aerosols. Nature. 2001, 409 (6821), 695-697. Jacobson, M. Z. Control of fossil-fuel particulate black carbon and organic matter, possibly the most effective method of slowing global warming. J. Geophys. Res. 2002, 107 (D19), 4410. Jacobson, M. Z. Correction to: “Control of fossil-fuel particulate black carbon and organic matter, possibly the most effective method of slowing global warming”. J. Geophys. Res. 2005, 107, (D14105). Kandlikar, M. Air pollution at a hotspot location in Delhi: Detecting trends, seasonal cycles and oscillations. Atmos. Environ. 2007, 41 (28), 5934-5947. Kandlikar, M. The relative role of trace gas emissions in greenhouse abatement policies. Energy Policy. 1995, 23 (10), 879-883. Kandlikar, M.; Ramachandran, G. The causes and consequences of particulate air pollution in urban India: a synthesis of the science. Annu. Rev. Energ. Env. 2000, 25, 629-684. Kathuria, V. Impact of CNG on vehicular pollution in Delhi: a note. Transport. Res. D-Transp. Environ. 2004, 9 (5), 409-417. Kojima, M. Breathing clean: Considering the switch to natural gas buses. World Bank: Washington, DC, 2001. Lipman, T. E.; Delucchi, M. A. Emissions of nitrous oxide and methane from conventional and alternative fuel motor vehicles. Climatic Change. 2002. 53 (4), 477-516. Mazzi, E. A.; Dowlatabadi, H. Air quality impacts of climate mitigation: UK policy and passenger vehicle choice. Environ. Sci. Technol. 2007, 41 (2), 387-392. McDonald, J. D.; Eide, I.; Seagrave, J.; Zielinska, B.; Whitney, K.; Lawson, D. R.; Mauderly, J. L. Relationship between composition and toxicity of motor vehicle emission samples. Environ. Health Perspect. 2004, 112 (15), 1527-1538. Mehta, R. History, politics and technology of CNG-diesel switch in Delhi, in Land Use, Transportation and the Environment. The Transport Asia Project: Pune, India, 2001: http://www.seas.harvard.edu/TransportAsia/.  59  Menon, S.; Hansen, J.; Nazarenko, L.; Luo, Y. Climate effects of black carbon aerosols in China and India. Science. 2002, 297 (5590), 2250-2253. Narain, U.; Krupnick, A. The impact of Delhi's CNG program on air quality. Resources for the Future: Washington, DC, 2007. Pope, C. A., III; Dockery, D. W. Health effects of fine particulate air pollution: Lines that connect. J. Air Waste Manag. Assoc. 2006, 56 (6), 709-42. Rabl, A. Environmental benefits of natural gas for buses. Transport. Res. D-Transp. Environ. 2002, 7 (6), 391-405. Ramanathan, V.; Ramana, M. V.; Roberts, G.; Kim, D.; Corrigan, C.; Chung, C.; Winker, D. Warming trends in Asia amplified by brown cloud solar absorption. Nature. 2007, 448 (7153), 575-578. Ravindra, K.; Wauters, E.; Tyagi, S.; Mor, S.; Van Grieken, R. Assessment of air quality after the implementation of compressed natural gas (CNG) as fuel in public transport in Delhi, India. Environ. Monit. Assess. 2006, 115 (1), 405-417. Schifter, I.; Diaz, L.; Lopez-Salinas, E.; Avalos, S. Potential impacts of compressed natural gas in the vehicular fleet of Mexico City. Environ. Sci. Technol. 2000, 34 (11), 2100-2104. Shine, K. P.; Berntsen, T. K.; Fuglestvedt, J. S.; Skeie, R. B.; Stuber, N. Comparing the climate effect of emissions of short- and long-lived climate agents. Philos. Trans. R. Soc. London, Ser. A. 2007, 365 (1856), 1903-1914. Swart, R.; Amann, M.; Raes, F.; Tuinstra, W. A good climate for clean air: Linkages between climate change and air pollution. Climatic Change. 2004, 66 (3), 263-269. Takeuchi, A.; Cropper, M. L.; Bento, A. The impact of policies to control motor vehicle emissions in Mumbai, India. J. Reg. Sci. 2007, 47 (1), 27-46. TERI, TERI Energy Data Directory and Yearbook 2004/05. The Energy and Resources Institute: New Delhi, India, 2006. West, J. J.; Fiore, A. M.; Naik, V.; Horowitz, L. W.; Schwarzkopf, M. D.; Mauzerall, D. L. Ozone air quality and radiative forcing consequences of changes in ozone precursor emissions. Geophys. Res. Lett. 2007, 34, L06806. UNFCCC. Clean Development Mechanism (CDM) Procedures. 2007: http://cdm.unfccc.int/Reference/Procedures/index.html.  60  3. Chapter 3: Determinants of PM and GHG emissions from natural gas fueled auto-rickshaws in Delhi7  3.1. Introduction  Auto-rickshaws are three-wheeled motor vehicles that operate as low-cost taxis and fill an important niche in the public transportation systems of many cities in developing countries. They have a simple, lightweight chassis with open sides, a canvas top and motorcycle-style engine and controls. There is a bench-style seat behind the driver with room for up to three passengers. Table 3.1 gives the specifications of the Bajaj auto-rickshaws in Delhi, which run on compressed natural gas (CNG). The current auto-rickshaw population in India is between 3 and 3.5 million, out of approximately 4.5 million globally. Since 2002, auto-rickshaws, buses and taxis in Delhi have been fueled by natural gas. This transition to ‘clean fuel’, mandated by a 1998 Supreme Court order (Mehta 2001), is a case study of an air-quality policy with potentially widespread technical and socio-economic implications. Effort has been expended to quantitatively evaluate the impact of this initiative on air quality (Narain and Krupnick 2007; Kandlikar 2007) as well as climate-forcing emissions (Reynolds and Kandlikar 2008). However, auto-rickshaws are the least studied motorized transportation mode, and there is substantial uncertainty regarding their activity, fuel consumption and emissions. In Delhi the number of auto-rickshaw registrations is capped at 83,000, but the Environment Pollution (Prevention & Control) Authority (EPCA) for Delhi believes that the number of vehicles is only around 55,000 (EPCA 2010). Auto-rickshaws are the focus of much attention because it has been claimed that a large proportion of the fleet emits ‘visible smoke’ (EPCA 2004), and there is concern about the adverse health effects of exposure to traffic-related air pollution (HEI 2010). In addition, motor vehicles are an important source of climate-forcing emissions such as CO2, CH4 and particulate matter.  7  A version of this chapter has been accepted for publication: Reynolds, C.C.O., Kandlikar, M., and Badami, M.G. “Determinants of PM and GHG emissions from natural gas fueled auto-rickshaws in Delhi”. Transportation Research Part D: Transport and Environment.  61  This paper describes a structured survey of auto-rickshaw drivers in Delhi to address knowledge gaps on the impacts of real-world auto-rickshaws. It was designed to elicit a range of information from drivers related to owning and operating a commercial auto-rickshaw: fuel use (and consequently CO2 emissions), vehicle activity, maintenance practices, technical challenges, and socio-economic factors. Additionally, observations of visible smoke and the presence of oily residue in the tailpipes of vehicles were recorded during the survey (and later checked against laboratory measurements) as indicators of particulate matter emissions and vehicle condition. Here we focus on survey questions related to vehicle activity and fuel consumption, data particularly important for energy and CO2 emissions inventories. We also present an analysis of driver- and vehicle-related variables that are associated with auto-rickshaws that emit high levels of particulate matter. Table 3.1. Specifications for spark-ignited auto-rickshaws fueled with compressed natural gas (CNG). Model / engine type a  Bajaj RE:CNG 4-stroke  Bajaj RE:CNG 2-stroke  Fuel type  Compressed natural gas  Compressed natural gas  (plus gasoline option for ‘limp-home’ mode) Fuel tank capacity  CNG: 4kg (~30 litre tank);  CNG: 4kg  Gasoline: 3 litre tank Displacement  173.5 cc (single-cylinder)  145.5 cc (single-cylinder)  Max power  4.8 kW @ 5000 rpm  6.0 kW @ 5000 rpm  Max torque  9.3 Nm @ 2500 rpm  11.5 Nm @ 4000 rpm  Gross vehicle weight  358 kg  305 kg  Maximum recommended payload  325 kg  325 kg  Maximum number of passengers  3  3  a  In addition to CNG, the other fueling options available for the spark-ignited engines are gasoline (called petrol in  India), and liquefied petroleum gas (LPG); there is also a diesel engine option (Bajaj Auto, 2010b).  3.2. Methods  3.2.1. Structured survey Structured surveys of a representative sample of 349 auto-rickshaw drivers in November– December 2008 and 32 in September 2009 were conducted in Delhi; the latter as a follow-up 62  during a study to conduct detailed emissions characterization tests for a sample of autorickshaws (Reynolds et al. 2009). Five people were hired to interview drivers using a structured survey instrument. The interviews were conducted at seven critical transportation nodal points around the city, which were identified in consultation with local transportation experts at the Transportation Research and Injury Prevention Programme at the Indian Institute of Technology, Delhi. Locations included a central commercial district at the heart of Delhi, two major traffic intersections, and four rail stations and/or inter-state bus terminals. These locations are all transportation hubs that are well-served by auto-rickshaws. By using dispersed sampling at major transportation nodal points, the likelihood of obtaining a representative sample of autorickshaw drivers from the overall population in Delhi was increased.8 As part of the survey protocol, interviewers conducted a two-part observation procedure to check for indications of high particulate matter emissions. The interviewers first conducted a visual inspection of the exhaust pipe for presence of oily residue, and second, for visible smoke at engine start-up. Vehicles were classified as being a ‘high-PM emitter’ if they met one of the above criteria. This method replicated the process that has been used by Delhi authorities to identify high-emitting auto-rickshaws (EPCA 2004).  3.2.2. Calibration of PM observations In addition to the survey, a sample of actual auto-rickshaws was brought to a state-of-the-art vehicle testing facility near Delhi and chassis-dynamometer testing was used to measure pollutant emission factors, including particulate matter (Reynolds et al., 2009). This sample was selected to ensure that adequate numbers of each age-group and engine type were tested, and included 17 vehicles with 4-stroke CNG engines and 11 with 2-stroke CNG engines. The drivers of these auto-rickshaws were also interviewed using the survey instrument described above. This made it possible to compare certain values calculated from the survey responses (such as fuel consumption) against the measured values, and also to test the observational procedure used in the broader sample to classify the auto-rickshaws in the lab-sample group as being ‘high-PM 8  Visual observations can be subjective. To minimize the uncertainty in classifying vehicles as high-PM emitters interviewers were trained at a two-day workshop prior to commencing data collection. Also, the temperature of the engine and the ambient air can affect the visibility of particulate matter emissions. Alternative, quantitative methods were considered, such as the use of simple smoke opacimeters, but were not possible due to the number and dispersion of measurements needed.  63  emitters’ (18 vehicles) or ‘low-PM emitters’ (9). Particulate matter emission factors, in units of milligrams of PM per kilogram of fuel consumed, were available for each vehicle. A nonparametric Mann–Whitney two-sample test confirmed that the sample classified as ‘high-PM emitters’ were drawn from a different population to ‘low-PM emitters’. This assessment confirms that our observational test functions as a reasonable indication of which vehicles are high-PM emitters, although it does not replace the need for actual emissions testing to quantify emission factors.  3.2.3. Determining correlation between survey variables and ‘high-PM emitters’ We estimated the effect of independent variables (Table 2) on the probability of observing high emitters using stepwise logistic regression. A generalized linear model with maximum likelihood estimation and a logic link function was used. Analysis was done using the R Statistical Computing package. The model was estimated for all vehicles pooled, then by engine type (i.e. grouping vehicles with 2-stroke and 4-stroke engines separately), and finally by ownership (i.e. grouping vehicles by whether they were operated by drivers who owned their vehicles – ‘ownerdrivers’ – or those who rented their vehicles – ‘renter-drivers’). Some variables were found not to be significant predictors of ‘high-PM emitters’ in any of the analyses, possibly partly due to missing data and insufficient resolution in drivers’ responses, and are indicated in Table 3.2.  64  Table 3.2. Variables collected in the survey and included in the logistic regression model. Dependent variables High-PM emitters  Vehicles were classified as ‘high-PM emitters’ if visible smoke was observed when the vehicle was started, and/or oil was observed coating the inside of the exhaust pipe. a  Independent variables – Demographic Driver Age b  Age of the auto-rickshaw driver in years  Driver Experience b  Total number of years driving (a) CNG-fueled auto-rickshaws and (b) any type of auto-rickshaws  Income  b  Take-home income (after expenses such as vehicle rent, fuel and maintenance) for the drivers’ previous day of work  Own vs. Rent  b  Differentiates drivers who own the vehicle they drive (‘owner-drivers’) from those who rent it from another person (‘renter-drivers’)  Rent Time  How long the driver of a rented auto-rickshaw has been renting this particular vehicle (renter-drivers only)  Bought New or Used  b  Whether owner-drivers bought their vehicle new or used.  Independent variables – Vehicle-related Vehicle Age  Age of the vehicle in years (where 2009 = 0, 2008 = 1, etc.)  Engine Type  Differentiates between vehicles with 2-stroke and 4-stroke engines  Breakdowns  Total number of vehicle breakdowns in the last month; note that this metric includes breakdowns that were not engine-related, so could only be considered a general measure of the condition of the vehicle  Fuel Consumption b  Vehicle’s fuel consumption, calculated from the distance traveled and fuel used during its previous day of work  a  This method was calibrated against a subset of vehicles, for which particulate matter emissions factors were  obtained during chassis-dynamometer testing (see section 2.2). b  These demographic determinants were not found to be significant predictors of the dependent variable in any of  the models.  3.3. Results  3.3.1. Fleet characteristics, vehicle activity and GHG emissions Vehicle ages in the survey data form a bimodal distribution, with modal peaks around the 2001 and 2007/2008 model years. Figure 3.1 gives the distribution of the two engine types present in 65  Delhi auto-rickshaws. Only 10% of the survey participants drove auto-rickshaws with 2-stroke engines, and all but two were model year 2000 or earlier. Although new 2-stroke CNG vehicles are presently available from Bajaj Auto (2010), their sale has not been permitted in Delhi since 2002 (EPCA 2004).  Figure 3.1. Vehicle model year frequency distribution by engine type (N = 349) The reason for the unusual age distribution is possibly due to the rapid transition from gasoline to CNG fuel. In 1998, the Supreme Court categorically ordered that all pre-1991 auto-rickshaws must be replaced with those running on ‘clean fuels’, which in effect meant compressed natural gas. Only 4-stroke auto-rickshaws fueled with compressed natural gas met the requirements of the Court for new auto-rickshaws, while conversions to CNG using engine ‘retrofit kits’ were allowed for model year 1991 vehicles and later. Widespread replacement of pre-1991 vehicles happened around the years 2000–2001, and it is likely to be the reason behind the first mode in the distribution. However, not all of the scrapped vehicles were replaced. During the transition to CNG, the number of registered auto-rickshaws in Delhi dropped from over 82,000 in 1997 (the maximum permitted number of registrations at the time) to 43,000 in 2002, and then rose to only 53,000 in 2004 (EPCA 2004; EPCA 2010). Lack of access to capital may have prevented many drivers whose vehicles were scrapped from purchasing new autos, despite government incentives. It appears that not very many new auto-rickshaws were added to the fleet between 2004 and 2006, but sales increased again in 2007 and appears to have continued at similar volumes to the present. Officially, no commercial vehicles older than 15 years are allowed to operate in the 66  Delhi metropolitan area (Mehta, 2001), which at the time of the original survey meant vehicles of model year 1993 and older should not have been present. Indeed, fewer than 1.5% of vehicles in the survey were in that age-category, which would suggest that the policy of restricting use of 1993 and older vehicles has largely been successful. The combined dataset (N = 381) was used to determine vehicle activity and fuel consumption (Table 3.3). The survey participants were asked to recall and report on their previous day of work. On average, drivers reported traveling approximately125 ± 50 km (reported uncertainty is plus/minus one standard deviation) each day during daytime working hours. Note that no significant difference was observed between the values reported by owner-drivers and renterdrivers. To make an estimate of average vehicle kilometers per day, it was necessary to take into account the proportion of renter-drivers who work on a shift basis, i.e. those whose vehicle was also driven at night (by a second, partner driver). 39% of the fleet fell into this category. If it is assumed that at night, the vehicles are driven half as far as during the daytime shift, it is estimated that each vehicle traveled an average of 150 ± 60 km per 24-h period. Therefore our best estimate of each auto-rickshaw’s annual vehicle kilometers traveled is ~54,000 km/year. Our findings are comparable to those reported by Moore (2007), who asked 95 auto-rickshaw drivers to estimate how far they traveled each day on average and reported a mean daily value of ~160 ± 70 km. Based on reported distance traveled and fuel purchase, the real-world fuel consumption of 4stroke and 2-stroke auto-rickshaws was calculated as being 3.51 and 4.25 kg/100 km respectively. The data suggest that auto-rickshaws with 2-stroke engines consume over 20% more fuel than their 4-stroke counterparts per kilometer, and this difference is statistically significant at the 1% level. It was not possible to compare these values to official fuel consumption figures based on standardized testing, because such data does not exist. Manufacturers are not presently required to report certified fuel consumption values in India. However, the fuel consumption estimates based on the survey data were about 50% higher than values estimated for a sub-set of vehicles during chassis-dynamometer testing over the regulatory Indian Drive Cycle. There are a number of plausible reasons for these differences: real-world driving is more fuel intensive than the Indian Drive Cycle, with more stop–start driving, longer idling time, more aggressive acceleration, and higher engine loading (due to 67  passengers’ weight); fuel consumption estimates based on only one day of fuel purchases (during which time the drivers fill up once or twice), rather than averaged over a longer period; and finally, drivers might not have accurately responded to the survey question about distance traveled per day because many of the vehicles’ odometers do not function after they are more than one or two years old. In contrast to the distance estimates, the drivers surveyed had very precise recall of how much money they had spent on fuel in their previous day of work. Table 3.3. Vehicle activity, fuel consumption and CO2 emissions for CNG-fueled autorickshaws in Delhi. 4-strokes  2-strokes  Significant difference?  (N = 330)  (N = 48)  (two-tailed t-test)  Number of trips with paying passenger  9.8 ± 4.6 a  10.5 ± 7.5  No  Distance traveled per day (km)  128 ± 48  119 ± 48  No  Fuel costs per day (Rs.)  Rs.79 ± 28  Rs.93 ± 30  Yes (p << 0.01)  3.96 ± 1.4  4.63 ± 1.5  Yes (p << 0.01)  Total distance traveled per day (km) c  153 ± 57  142 ± 57  No (mean: 147 ± 57km)  Fuel consumption (kg/100km)  3.51 ± 2.5  4.25 ± 1.7  Yes (p < 0.01)  98 ± 68  118 ± 47  Yes (p < 0.01)  5.44 ± 2.7  6.13 ± 1.0  Yes (p < 0.01)  50,000  5,000  -  2790 ± 1000  260 ± 100  -  270 ± 100  30 ± 10  -  Per driver (daytime):  Fuel use per day (kg)  b  Per vehicle:  CO2 emissions (g/km)  d  Annual CO2 emissions (tonnes)  d  Delhi fleet, annual: Approximate number of vehicles e Annual fleet distance (million km) Annual CO2 emissions (thousand tonnes)  d  a  Reported uncertainty in this table is plus/minus one standard deviation.  b  Calculated based on a fuel-price (at time of survey) of approximately Rs.20 per kg CNG.  c  This estimate takes into account that 40% of the renter-drivers (who make up 68% of both 4-stroke and 2-stroke  drivers) work on a shift basis, so their vehicles (27% of the total fleet) would also be driven at night. We assume that the distance traveled at night is 50% of the distance they travel during the day. d  Assuming natural gas fuel is completely burned. This is a conservative value because there is always some  unburned methane released (which has a global warming potential 25 times that of CO2). e  Assuming a fleet of approximately 55,000 vehicles, of which around 10% have 2-stroke engines.  68  Despite the uncertainty evident in the figures quoted above, the calculated fuel consumption (and consequently the CO2 emissions) from three-wheeled auto-rickshaws, with their very simple single-cylinder engines, are similar or lower than the most efficient small automobiles. For example, the Maruti 800 has been a particularly popular small car in India for over 2 decades, in part because of its fuel efficiency. This inexpensive vehicle has a 3-cylinder 800 cc engine, and users report that its real-world fuel consumption averages 15.7 km/l (Pundir 2008), which is energetically equivalent to about 4.1 kg natural gas per 100 km. The higher fuel consumption of 2-stroke CNG engines means that they emit more CO2 than CNG 4-strokes. Assuming that all of the fuel is converted to carbon dioxide, and using the survey values for fuel consumption and vehicle kilometers traveled, each 2-stroke auto-rickshaw in Delhi can be expected to produce 6.1 ± 1.0 tonnes of CO2 per year, compared to 5.4 ± 2.7 tonnes annually for 4-stroke vehicles. However real-world 2-stroke engines fueled with natural gas emit significant quantities of methane during the exhaust scavenging process. Since methane is a potent greenhouse gas with a global warming potential 25 times that of carbon dioxide (Forster et al. 2007), the CO2-equivalent emissions from 2-stroke CNG engines is likely to be significantly higher than the direct CO2 emissions reported here.  3.3.2. Classifying auto-rickshaws as ‘high-PM emitters’ The survey data provide strong evidence that 2-stroke CNG auto-rickshaws are more likely to emit visible smoke than 4-strokes. Out of 332 vehicles with 4-stroke engines and 49 with 2stroke engines, 82% of the auto-rickshaws with 2-stroke engines were classified as ‘high-PM emitters’, compared to only 13% of those with 4–stroke engines. The conditional probability of observing an equal number of smoky 4-strokes as 2-strokes, given that 90% of vehicles in the Delhi fleet have 4-stroke engines, could lead to the misconception that the two different engines types are each as likely to be high-emitters. In October 2004, the EPCA and the Transport Department of Delhi conducted ‘a special drive against three-wheelers and other vehicles, which emit visible smoke’ (EPCA 2004). About 50% of the 168 three-wheelers caught emitting smoke had 4-stroke engines. Bajaj Auto, the manufacturer of three-seater auto-rickshaws, was requested to submit a report to EPCA on the possible causes of visible smoke emissions from 4-strokes, and they concluded that the main 69  cause of visible smoke emission was ‘lack of proper maintenance of the vehicle as per recommended schedules and use of sub standard change parts’ (EPCA 2004). In the present study, a logistic model was used to explore the influence of a wider range of demographic and vehicle-related variables on whether or not an auto-rickshaw would be classified as a ‘high-PM emitter’. This type of model allows the effect of a given determinant – say vehicle age – to be examined while controlling for other factors. The results of the analyses are presented in Table 3.4. The model was estimated with five datasets: the first included all vehicles; the second and third examined 4-stroke and 2-stroke vehicles as separate groups; and the 4th and 5th examined the groups of owner-drivers and renter-drivers separately.  Analyzed  Size (N)  Own vs. Rent  Rent Time  Vehicle Age  Engine Type  Breakdowns  Fuel  1. All autos  376  -2.34 ***  -0.49  n/a  0.18 ***  2.64 ***  -0.08  -0.05  2. 4-stroke engines  328  -2.55 ***  -0.35  n/a  0.23 ***  n/a  -0.02  -0.13  3. 2-stroke engines  48  2.51  -0.83  n/a  -0.13  n/a  -0.34  0.39  4. Owner-drivers  116  -1.19  n/a  n/a  0.12  3.64 ***  0.05  -0.36  5. Renter-drivers  259  -3.91 ***  n/a  0.47 *  0.25 ***  2.39 ***  -0.35 *  0.01  Consumption  Sample  Constant  Sample  Regression  Table 3.4. Results of logistic regression analyses, according to the different groupings analyzed.  n/a = not applicable (variable not included in analysis) Significance codes: *** p<0.01, ** p<0.05, * p<0.1 Dependent variable in each case was whether or not the auto-rickshaw was classified as a ‘high-PM emitter’ (i.e. observation of oily tailpipe and/or visible smoke on start-up).  In the analysis of all vehicles, engine type and vehicle age were both strong predictors (significant at 1%) of whether or not a vehicle would be classified as a ‘high-PM emitter’. Changing the engine type from 4-stroke to 2-stoke increases the odds ratio that a given vehicle will be found to be a high-PM emitter by a factor of 2.6. For vehicle age, the odds ratio is increased by a factor of 0.18 for each additional year the vehicle has been driven. In other words a vehicle that is 10 years older than another will have its odds ratio of be a high-PM emitter increased by 1.8, all other variables being held constant. To investigate if any determinants  70  become significant within subgroups in the dataset, and to mitigate errors due to potential correlation between engine type and age, we examined various subgroups separately. In the analysis of vehicles with 4-stroke engines, only vehicle age was found significant. Each additional year of age increased the odds that a 4-stroke auto-rickshaw would be a high-PM emitter by a factor of 0.23. For the analysis of the 2-stroke group, no variables are statistically significant, probably because of the relatively small sample size and the limited range of model years for 2-stroke vehicles in Delhi (mostly 1998-2000). However, this does not mean that age is not an important factor for 2-strokes. Since about 80% of the 2-stroke vehicles are high-PM emitters, despite running on ‘clean’ compressed natural gas, it is reasonable to conclude that the program to phase them out has produced environmental benefits. Overall, engine type is the most important predictor of whether a vehicle will be a high-emitter; the odds of this being then case increase by 2.4-3.6 for 2-stroke compared to 4-stroke vehicles, depending on the group examined. This is because 2-stroke engines inject oil into the fuel–air mixture for lubrication of the moving parts. Smoke from 2-strokes is also exacerbated by drivers using the incorrect type of oil, often 4-stroke crank-case oil instead of ‘low-smoke’ 2-stroke oil, to reduce costs, and possibly the use of excessive amounts of oil (Badami and Iyer 2006). In contrast, visible smoke from 4-stroke engines was probably due to lubricating oil leaking past worn piston rings or valve seals into the combustion chamber (EPCA 2004). The findings suggest that the carbureted 2-stroke engines currently in use in Delhi are highly polluting and should be phased out, regardless of the fact that they use a clean fuel such as CNG. In Delhi, PM emissions will be reduced as the remaining conventional 2-stroke auto-rickshaws are scrapped when they reach the end of their useful lives, and replaced with new 4-stroke CNG auto-rickshaws. In other Asian cities with a high population of 2-stroke auto-rickshaws, where CNG or LPG (liquefied petroleum gas) is used, the transition to 4-stroke engines should not be delayed by the mistaken belief that the clean fuel somehow makes conventional 2-stroke engine a ‘clean technology’.  3.4. Conclusions  71  The natural gas-fueled auto-rickshaw fleet in Delhi is composed primarily of vehicles with 4stroke engines. 10% of the vehicles have 2-stroke engines, which have about 20% higher fuel consumption with associated greater CO2 emissions. To identify auto-rickshaws that have the potential to be high-emitters of particulate matter, a simple observational procedure was used, and the auto-rickshaws with 2-stroke engines were found to be far more likely to be high-PM emitters. Within the group of 4-stroke vehicles, age was found to be a significant predictor of high-PM emitters. Given concerns about the adverse health effects of exposure to particulate matter emissions, phase-out of conventional 2-stroke engine technology appears to be a desirable policy for CNG auto-rickshaws in Delhi.  72  3.5. References Badami, M. G.; Iyer, N. V. Motorized two-wheeled vehicle emissions in India - Behavioral and institutional issues. Transp. Res. Rec. 2006, 1954, 22-28. Bajaj. Commercial Vehicles, 2010: www.bajajauto.com/commercial_vehicle.asp. DOT. Pollution Under Control (PUC) Norms. Department of Transport, Government of Uttarakhand: India, 2010: http://gov.ua.nic.in/transport/pollution_under_control.htm. EPCA. Report on the increase in the number of three-wheelers in Delhi. Report No. 9. Environment Pollution (Prevention & Control) Authority for the National Capital Region: Delhi, India, 2004. EPCA. Review of existing cap on the number of three-wheelers in Delhi and its implications for pollution and congestion. Report No. 34. Environment Pollution (Prevention & Control) Authority for the National Capital Region: Delhi, India, 2010. Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.; Haywood, J.; Lean, J.; Lowe, D.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Dorland, R. V. Changes in atmospheric constituents and in radiative forcing, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon, S. et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, NY, 2007. HEI. Traffic-Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Special Report 17; Health Effects Institute: Boston, MA, 2010. Kandlikar, M. Air pollution at a hotspot location in Delhi: Detecting trends, seasonal cycles and oscillations. Atmos. Environ. 2007, 41 (28), 5934-5947. Lorenz, N.; Bauer, T.; Willson, B. Design of a direct injection retrofit kit for small two-stroke engines. SAE Tech. Pap. 2005, 2005-32-0095. Mehta, R. History, politics and technology of CNG-diesel switch in Delhi, in Land Use, Transportation and the Environment. The Transport Asia Project: Pune, India. 2001: http://www.seas.harvard.edu/TransportAsia/Dec01Papers.htm. Moore, S. The Sustainability of Auto-Rickshaws in Delhi: Environmental, Economic and Social Perspectives. MASc Thesis, Centre of Environmental Policy, University of London: London, UK, 2007. Narain, U.; Krupnick, A. The impact of Delhi's CNG program on air quality. RFF DP 07-06. Resources for the Future: Washington, DC, 2007.  73  Pundir, B. P. Fuel Economy of Indian Passenger Vehicles: Status of Technology and Potential FE Improvements. Report for Greenpeace India Society: Bangalore, India, 2008: http://www.iitk.ac.in/mech/New_Books/books.htm. Reynolds, C. C. O.; Grieshop, A. P.; Kandlikar, M. Testing Emissions from auto-rickshaws to inform better air pollution control policies. SIM-air Working Paper Series, No. 28-2009. 2009: http://www.urbanemissions.info/simair/simseries.html. Reynolds, C. C. O.; Kandlikar, M. Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860-5865.  74  4. Chapter 4: Climate and health relevant emissions from inuse Indian three-wheelers fueled by natural gas and gasoline9  4.1. Introduction  Motor vehicle emissions make a large contribution to the urban air pollution in Asian cities (Kim Oanh et al. 2006; Hopke et al. 2008), and population exposure to mobile sources is exacerbated by the proximity of individuals, residences and workplaces to traffic (HEI 2010). Transport is also a major contributor to climate change emissions, and policies aiming to improve air quality (AQ) may have climate impacts, and vice versa (Mazzi and Dowlatabadi 2007; Reynolds and Kandlikar 2008; Bell et al. 2008). CO, non-methane hydrocarbons (NMHC) and NOX are typically regulated for new vehicles, due to their health impacts (HEI 2010). Fine particulate matter (PM2.5) from spark-ignition engines is not regulated in most jurisdictions. However, multiple lines of evidence have connected PM2.5 exposure to adverse health affects (Pope and Dockery 2006). PM2.5 components also affects regional and global climate by changing the atmospheric radiation balance (Grieshop et al. 2009). Certain exhaust constituents, such as CO2 and CH4, are not regulated and are not directly hazardous to human health, but are critical greenhouse gases (Forster et al. 2007). Since transportation emissions can impact AQ and climate change, both aspects need to be considered when conducting emission measurement studies or implementing emission control policies. In India, the main policy approaches that have been used to address traffic-related air pollution are: (i) new vehicle emission standards (Bharat Stage Norms); (ii) an inspection and maintenance program for in-use vehicles (‘Pollution Under Control’, PUC); (iii) limiting heavy-duty diesel truck activity to night-time hours; and (iv) use of alternative fuels – such as compressed natural gas (CNG) and liquefied petroleum gas (LPG) – to replace gasoline and diesel. The Bharat Stage Norms and PUC are described in Appendix C. In 1998, the Indian Supreme Court ordered that 9  This chapter has been submitted: Reynolds, C.C.O., Grieshop, A.G. and Kandlikar, M. “Climate and health relevant emissions from in-use Indian three-wheelers fueled by natural gas and gasoline”.  75  all public transportation vehicles in Delhi, including three-wheeled passenger carriers or autorickshaws (‘tuktuks’ in other parts of the world), must operate on ‘clean fuel’ (Mehta 2001). As a result, all buses, taxis and auto-rickshaws in Delhi were retrofitted to run on CNG by 2003. Since then, new vehicles with CNG engines have been available so less retrofitting has occurred. A number of other jurisdictions in India and other countries, notably Pakistan and Bangladesh in Asia, but also Latin American and European countries, have switched or are considering switching to CNG-fueled vehicles, citing environmental considerations (Yeh 2007). The impact of Delhi’s CNG policy on both AQ and climate has been studied in a number of ways. Time-series analyses of ambient AQ data (a ‘top-down’ approach) have shown mixed results: a spectral analysis of pollution measurements at a traffic ‘hot-spot’ (Kandlikar 2007) noted that an abrupt decrease in CO coincided with the CNG switch in 2002. Meanwhile the observed trend in NOX (gradual increase until 2004, followed by a small decline) was likely related to other changes in Delhi’s vehicular traffic, including increasing number of vehicles that were built to meet more stringent emission standards and changes in the fleet composition. The study also suggests that changes in PM10 concentrations were most likely unrelated to traffic sources. Another study found significant correlation between CNG use in buses and reduced ambient SOX and PM10 concentrations, but found no evidence of positive benefits of a switch to CNG for auto-rickshaws and taxis (Narain and Krupnick 2007). An alternative, ‘bottom-up’ approach to assess emission mitigation options uses emission inventories based on emission factors (EFs: pollutants emitted per fuel consumed or vehicle activity), fleet composition and activity levels. For example, two of us recently evaluated the climate impact of the CNG-switch in Delhi (Reynolds and Kandlikar 2008) using our best estimates of EFs. However, lack of empirical emission data for in-use vehicles in developing countries is a source of significant uncertainty in such bottom-up studies. Vehicle emissions can range widely in different settings due to variations in fuel quality, uncertain vehicle age distributions, and non-existent or ineffective inspection and maintenance programs leading to a fleet with a high fraction of poorly-tuned or malfunctioning engines (Bond et al. 2004). A recent inventory of road transportation emissions in India reports that uncertainty in fleet average EFs was lowest for NOX (about ±20%), and greatest for CO (-51% to +68%) (Baidya and Borken-  76  Kleefeld 2009). Uncertainty in PM2.5 emissions from spark-ignition engines may be even greater (Bond et al. 2004). Furthermore, because CNG and LPG use is less prevalent than conventional transportation fuels, few EFs for vehicles using those fuels are available. The Automotive Research Association of India (ARAI) conducted a measurement campaign to develop EFs for a wide range of in-use Indian vehicles, including those fueled with CNG (ARAI 2007). However, the study examined only one or two test vehicle for each engine/alternative fuel combination (each tested before and after maintenance), which did not allow for an assessment of inter-vehicle variability. Vehicle emissions studies using larger sample sizes suggest that as many as 25 vehicles may be needed to estimate EFs, given normal variability in a fleet (Kuhns et al. 2004; Subramanian et al. 2009). This study measured a comprehensive set of emissions relevant to AQ (PM2.5, CO, THC, NOX) and climate change (CO2, CH4, organic carbon [OC], elemental carbon [EC], CO and NMHC) from a sample of in-use auto-rickshaws recruited from the on-road fleet in Delhi. Autorickshaws from Delhi were chosen for the study because: (a) they operate on CNG, (b) they are ‘for-hire’ vehicles, and are hence readily recruited, and (c) they all have the same basic chassis design and function (manufactured by one company, but available with different engine types and a wide range of model years). In addition, there are more than 55,000 CNG auto-rickshaws for hire in Delhi (EPCA 2010), which is about twice the number of conventional taxis. Autorickshaws provide flexible mobility that is faster and more reliable than city buses while costing less than taxis (Iyer 2003). There are an estimated 3 million auto-rickshaws in India, around twothirds of the world total (Iyer 2003; Bajaj 2010). The overall aims of this study were to: (a) determine emission factors from light-duty CNGfueled engines for use in inventories of health- and climate-relevant emissions, (b) compare emissions from 2-stroke and 4-stroke CNG engines, (c) assess the extent to which emissions from in-use CNG vehicle vary and exceed Bharat Norms for new vehicles, and (d) explore the influence of fuel choice and vehicle age on emissions from a subset of the vehicles.  77  4.2. Experimental methods  4.2.1. Vehicles and fuels Emission testing was conducted on in-use CNG-fueled auto-rickshaws, recruited from transportation nodes (such as bus-stations) in Delhi and nearby Gurgaon. The test vehicles were powered with 4-stroke (CNG-4S, N = 17) and 2-stroke (CNG-2S, N = 13) spark-ignition engines. Of the 4-stroke vehicles, a subset also had functioning gasoline (petrol) fuel systems (PET-4S, N =11). Although most 4-stroke CNG auto-rickshaws are equipped with a ‘back-up’ gasoline fuel system, gasoline is almost twice as expensive as CNG in Delhi and is rarely used by operators. As a result, engines in back-up mode may not have functioned as well as those in dedicated gasoline engines. One of the dual-fuel vehicles was nearly new (denoted V14; less than 1000km on odometer), offering an opportunity to compare its emissions with more heavilyused vehicles and with Bharat Stage Norms. The 4-stroke vehicles ranged in model year (MY) from 2000 to 2009; our vehicle sample was bi-modally distributed with vehicles falling into a ‘new’ group (MY 2007-2009) and an ‘old’ group (MY 2000-2001). The 2-stroke vehicles were older (MY 1998-2001) and were thus treated as a single group in the analysis. None of the vehicles recruited for the study were of model year 2002-2005, which reflects the age distribution of Delhi auto-rickshaws (see Chapter 3). Table 4.1 gives the engine/fuel test-matrix for this study; full vehicle specifications are listed in Table C.2 (Appendix C). Table 4.1 Characteristics of vehicle groups in chassis dynamometer testing. Group  Fuel Type  Engine Type a  Number of vehicles and model years  CNG-4S  CNG  4-stroke, spark-  N=17 (‘New’: 2007-2009, N=9; ‘Old’:  ignition  2000-2001, N=8)  2-stroke, spark-  N=13 (‘New’: 2007-2009, N=7; ‘Old’: 2001,  ignition  N=4)  Gasoline  4-stroke, spark-  N=11 (1998-2001) b  (petrol)  ignition  CNG-2S  PET-4S  a  CNG  All vehicles were in-use three-wheeled auto-rickshaws with single-cylinder, spark-ignition engines manufactured  by Bajaj Auto Ltd. They were recruited from the commercial fleet in Delhi, and tested on the Indian Drive Cycle. b  The PET-4S group was a subset of the CNG-4S group operable on a ‘back-up’ gasoline fuel system.  78  Since the purpose of the study was to obtain ‘real-world’ emission factors, the vehicle operators were requested to bring their vehicles with full CNG and/or gasoline fuel tanks, with fuel from their regular fuel stations. Engine oil was not changed prior to testing, and the engines were not modified or tuned-up in any way. The vehicles were inspected for safety (e.g. brake function) and leaks (oil and exhaust system) and repaired if necessary to allow safe operation on the chassis dynamometer.  4.2.2. Data collection and analysis The testing was conducted in a vehicle test cell at the International Center for Automotive Technology (ICAT) in Manesar, near New Delhi (ICAT 2010). Vehicles were mounted on a chassis dynamometer, and driven over the Indian Drive Cycle (IDC); see Figs. C.1 and Table C.3 in Appendix C. The exhaust stream was diluted in a constant volume sampler; a bag-sample collected over the cycle was analyzed for gaseous emissions with an AVL-Pierburg AMA–4000 analyzer bench. Meanwhile PM2.5 was collected on a Teflon filter train (for mass measurement) and two quartz fiber filters (one in parallel with the Teflon filter and one behind the Teflon filter to correct for positive sampling artifact). Offline analyses yielded PM2.5, OC and EC mass concentration measurements. Details of the test protocol and emission measurements can be found in Appendix C. Fuel-based emission factors (g kg-1) were calculated from distance-based emission factors and fuel consumption estimates (SI). Fuel-based, rather than distance-based, emission factors are presented here because they are independent of fuel consumption, and hence may be used as estimates for other vehicles with similar engines but different mass or engine displacement (Singer and Harley 1996). Distance-based emission factors are used in the text where appropriate (e.g. to compare to emission standards and previous measurements), and can be readily calculated for individual tests by dividing the fuel-based emission factors by fuel consumption (both included in Table C.4, Appendix C). The fuel-based ‘global warming commitment’ (GWC, in g CO2-equivalent per kg fuel, g-CO2-eq kg-1), were calculated for each vehicle test by multiplying exhaust constituents by their 100-year global warming potential (GWP100) (Smith et al. 2000). Of the exhaust constituents measured 79  here, only CO2 and CH4 are included in the Kyoto Protocol. CO, NMHC, OC and EC also have substantial climate impacts and GWP100 estimates are available for them (Table C.5, Appendix C), but these values are highly uncertain so GWC thus quantified are not ‘tradable’ CO2equivalent emissions. We therefore report fuel-based GWC with Kyoto emissions only (GWCKyoto) as well as GWC with all global warming exhaust constituents (GWC-All). To compare emission factors for different groups, 95% confidence intervals (95% CI) were calculated by bootstrap resampling; hence statistically significant differences were defined as occurring at the 95% confidence level. For each group of N vehicles, 1000 random samples of size N were drawn with replacement. The resulting sample distribution gives an estimate for the original sample CI.  4.3. Results and discussion  4.3.1. Emission rates and fuel consumption Fuel-based emission factors for gaseous exhaust constituents are summarized in Figure 4.1. Measurements of vehicle fuel consumption, PM EFs and global warming commitment are summarized in Figure 4.2. Data for the 41 individual vehicle tests are listed in Table C.4 (Appendix C).  80  Figure 4.1 Fuel-based emission factors for gaseous pollutants, for 4-stroke CNG and gasoline/petrol-fueled auto-rickshaws (CNG-4S and PET-4S) and for 2-stroke CNG-fueled autorickshaws (CNG-2S): A: carbon monoxide, B: total hydrocarbons, C: methane, D: oxides of nitrogen, and E: carbon dioxide, with lines indicating upper limits on CO2 (for ideal combustion). Tukey box plots are used: the line in the box indicates group median, box ends give the interquartile range (IQR) and whiskers show data within 1.5 * IQR of the box end; outliers beyond this range are indicated by data points. NOX was the only species with EFs found to differ significantly (95% confidence interval) between ‘new’ (MY 2007-2009) and ‘old’ (MY 2000-2001) vehicle age groups. PET-4S vehicles have therefore been disaggregated by age in panel D; data for the PET-4S ‘old’ group are shown as individual points because of the small sample size (N=4).  81  Figure 4.2 Fuel-based particulate matter emission factors (A: PM2.5 mass, B: organic carbon fraction [OC/PM], C: elemental carbon fraction [EC/PM]), global warming commitment (D: GWC-Kyoto, E: GWC-All), and F: fuel consumption. PM2.5 emissions from CNG-2S vehicles are an order of magnitude larger than from the 4-stroke groups, so the data have been scaled by a factor of ten (Panel A). See Fig. 4.1 caption for box plot description.  Figures 4.1 and 4.2 show that most species have very wide inter-vehicle variability within groups, confirming the value of multiple tests for developing fleet emission factors. For example, THC emissions from CNG-4S vary by a factor of two (83 g kg-1 [95% CI: 54-116]), and PM2.5 for CNG-4S and CNG-2S data have coefficient of variation (COV) of greater than 1.0 (1.2 and 1.4, respectively). Despite the wide variation within vehicle types, substantial differences in emission factors were still found between vehicle categories, for example mean CH4 EFs (Fig. 4.1C) from CNG-2S (310 g kg-1 [281-344]) is significantly different from CNG-4S (50 g kg-1 [38-62]). Tables C.6 (gaseous emissions) and C.7 (fuel consumption, PM and GWC) in Appendix C list the mean and 95% CI of emission factors and fuel consumption data separated by fuel/engine-type group, and also by age for the 4-stroke engine groups. 82  When the 4-stroke auto-rickshaws were separated into ‘new’ and ‘old’ groups and their emissions compared by fuel type, no significant difference was observed for any exhaust constituents except for NOX from the gasoline-fueled vehicles. NOX emissions from the four old PET-4S auto-rickshaws were lower by almost a factor of 3, while CO emission rates of these vehicles were almost double those from newer PET-4S vehicles, though this difference is not significant at the 95% level (Figs. 4.1A and 4.1D). These differences are consistent with older vehicles running fuel-rich either due to poor fuel system condition or because operators and mechanics purposefully tune them thus to improve engine power and combustion reliability at the expense of fuel consumption (Heywood 1988). The fuel consumption for PET-4S (3.8 kg 100km-1) was in fact almost double that of CNG-4S group (2.1 kg 100km-1) despite the fact that the two fuels have very similar energy densities. Overall, mean CO and THC emissions for PET4S are about a factor of 10 and a factor of 3 higher than for CNG-4S, which is indicative of poor combustion efficiency in gasoline fueling mode. These findings are not surprising considering that the gasoline fuel systems in the PET-4S test vehicles were infrequently used. To highlight the importance of measuring a population of vehicles, we compared our emission measurements to those measured in other laboratory studies, such as that conducted by the Automotive Research Association of India (ARAI 2007). The ARAI study tested one or two vehicles per vehicle class (Table C.8, Appendix C), and extrapolating those results from a few tests to a larger population could be misleading. The two gasoline-fueled, 4-stroke autorickshaws tested by ARAI had 15 times lower CO and 8 times lower THC emissions than our PET-4S sample. ARAI’s results for CNG-fueled vehicles were based on only one 4-stroke and two 2-stroke vehicles. Emissions for the CNG 4-stroke auto-rickshaw (Bajaj Auto Ltd., MY >2000, tested after basic maintenance) are mostly within one standard deviation of our EF estimates. The exception is ARAI’s estimate for THC, which is one fifth of this study’s estimate (perhaps due to low CH4 emissions for the ARAI vehicle, though that was not reported). In contrast, ARAI’s study significantly underestimates emissions for retrofitted 2-stroke CNG autorickshaws compared to this study (based on two Bajaj Auto Ltd. vehicles, one MY <2000 and one MY >2000, each tested before and after maintenance). Our average CO, THC and PM2.5 EFs are more than a factor of 3 higher than those measured by ARAI, while our NOX EF is a quarter of ARAI’s (See Table C.8 in Appendix C for details). 83  In this study, we place emphasis on PM2.5 emissions because of their robust association with adverse health impacts, and because the composition of PM has a strong influence on its climate impacts. There is relatively little information about particulate emissions from CNG-fueled spark-ignition engines in the literature. To date, most epidemiological studies of the health impacts of exposure to PM2.5 have focused on mass (Pope and Dockery 2006), which we report here. There is some evidence that exposure to OC and EC in PM2.5 is associated with risk of cardiovascular disease (Peng et al. 2009), though the extent to which adverse health effects are attributable to these or other constituents of PM is still an open question (Brook et al. 2010). OC and EC fractions of PM2.5 are used in source apportionment studies and for calculating overall radiative forcing of PM2.5 from a given vehicle type, but we are unaware of any published values for CNG-fueled vehicles. CNG-4S emissions had OC/PM and EC/PM of 0.52±0.2 and 0.21±0.1, respectively. OC/PM from CNG-2S was 0.75±0.04, while EC from the vehicles in this group was below detection limits. The PET-4S vehicles had OC/PM and EC/PM of 0.49±0.3 and 0.37±0.22, respectively. Assuming an organic matter (OM) to OC ratio of 1.2 (Russell 2003), the carbonaceous components (OM plus EC) account for 81±14%, 90±5% and 82±16% of PM2.5 mass for CNG-4S, CNG-2S and PET-4S, respectively.  4.3.2. Real-world emissions vs. norms The nearly new auto-rickshaw (vehicle V14, 2009 model year, 4-stroke) was tested with both CNG and gasoline. It almost passed all Bharat Stage-II regulatory Norms (2009, see Table C.1 in Appendix C) in both fueling modes, except that it exceeded the THC+NOX standard in CNG mode by 10%. For model year 2010 and later, auto-rickshaw manufacturers must meet the more stringent Stage-III Norms, implying that emission controls will need to be improved. Gasolinefueled auto-rickshaws (two- and four-stoke) have been sold with oxidation catalysts since the 2000 model year but CNG-fueled auto-rickshaws have not yet needed a catalyst to meet the regulations. In India, new vehicles’ emissions are durability tested under laboratory conditions (30,000 km for Indian two- and three-wheelers), so regulators and policy-makers may assume that vehicles will continue to meet the emission Norms for several years. Based on our sample, this appears to be a reasonable assumption for vehicles with 4-stroke CNG engines, but emissions from vehicles 84  similar to our CNG-2S and PET-4S samples could increase significantly over their lifetime of a decade or more. Comparing average emission rates from the test vehicle groups to Bharat Stage-II Norms shows that while substantial increases in emissions can occur over time, deterioration doesn’t affect all fuel and engine types equally. CNG-4S vehicles fared the best, with mean CO (1.9±1.6 g km-1) and THC+NOX (1.7±1.0 g km-1) emission rates that remained within the Norms. In contrast, the PET-4S group exceeded the Norms by a factor of 15 and 3.5 for CO (32.7±16 g km-1) and THC+NOX (6.8±5.5 g km-1), respectively. However, we caution that the PET-4S EFs in this study may be biased high due to infrequent operation with gasoline fuel: almost half of the fuel carbon is converted to CO. The CNG-2S group had relatively low CO emissions (2.1±2.3 g km1  ), but THC+NOX (8.2±7 g km-1) exceeded the Norms by a factor of 4. Over 95% of THC+NOX  mass emissions from PET-4S and CNG-2S were THC. Using these EFs, the switch from gasoline to CNG in 4-stroke vehicles would appear to have resulted in a very large reduction in CO and THC.  4.3.3. Climate impacts of CNG Figures 2D shows GWC with only Kyoto constituents: GWC-Kyoto for CNG-2S (9710 g-CO2eq kg-1 [9000-10,500]) is more than double that of CNG-4S (3880 g-CO2-eq kg-1 [3600-4200]) and 5 times that of PET-4S (1840 g-CO2-eq kg-1 [1600-2200]). The climate impact of CNG vehicles is strongly dependent on CH4 emissions, which contribute 32% of GWC-Kyoto for CNG-4S and 80% of GWC-Kyoto for CNG-2S. Adding the other climate forcing constituents (CO, NMHC, OC and EC) does not significantly affect the calculation for the CNG vehicles: GWC-All is within 10% of GWC-Kyoto for both groups (Figs. 4.2D and 4.2E). For PET-4S, however, including the other climate forcing constituents almost doubles GWC-All compared to GWC-Kyoto, bringing it almost level with GWC-All for CNG-4S (~4100 g-CO2-eq kg-1). This is primarily due to the extremely high CO and NMHC emissions from PET-4S vehicles tested in this study. The implication of this analysis is that it should be sufficient for policy-makers to measure only Kyoto constituents when calculating GWC for spark-ignition CNG vehicles, due to the relatively low contribution of other exhaust constituents. Poorly maintained gasoline-fueled vehicles may have 50% or more of their GWC-All attributable to CO and NMHC emissions, an effect that merits further attention in future studies. 85  In our previous assessment of the climate impacts of the CNG switch in Delhi (Reynolds and Kandlikar 2008), GWC-All for auto-rickshaws was estimated based on emission factors from the literature: a fleet-average value of 113 g km-1 was used (engine-specific data were unavailable). Here we show that distance-based EFs for in-use CNG auto-rickshaws are highly dependent on engine type, with GWC-all for CNG-4S and CNG-2S of 89±16 g km-1 and 240±40 g km-1, respectively. Based on these EFs, the revised fleet average GWC-All for the Delhi CNG autorickshaw fleet (assuming 10% 2-stroke engines) would be 104±15 g km-1, thus our previous estimate was reasonable.  4.3.4. CNG engines: 4-stroke good, 2-stroke bad? To compare different engine-types and get a complete picture of their health and climate coimpacts, both regulated (CO, THC, NOX) and unregulated (PM2.5, CH4, CO2) species must be considered. Figure 4.1 shows that both two- and 4-stroke engine types produced similar levels of CO (Fig. 4.1A). 2-stroke CNG engines emitted five times more THC than 4-strokes (Fig. 4.1B), of which most is CH4 from unburned fuel (Figure 4.1C). Although CH4 does not directly cause adverse health effects, it is an important precursor to trophospheric ozone (West et al. 2006) and it is a strong greenhouse gas. In contrast, THC emissions from gasoline engines were mostly comprised of NMHC, many of which are air toxics and also contribute to regional ozone formation. 2-stroke engines’ NOX emissions were an order of magnitude lower than NOX from 4-stroke engines (Fig. 4.1D). Fuel consumption for CNG-2S vehicles (2.5 kg 100km-1 [95% CI: 2.4-2.7]) was about 20% higher than for the CNG-4S group; this is expected for 2-stroke engines due to scavenging losses (Faiz et al. 2004; Kojima et al. 2002). With respect to gaseous emissions, both engine types present tradeoffs. When PM2.5 emissions are examined, it becomes apparent that the legitimacy of CNG’s designation as a ‘clean fuel’ in spark-ignition engines relative to gasoline is strongly dependant on which engine technology is used. Figure 4.2A shows that CNG-2S vehicles emitted approximately 30 times more PM2.5 (14.2 g kg-1 [6.2-26.7]) than the CNG-4S group (0.5 g kg-1 [0.3-0.9]), on average. OC/EC analysis revealed that the OC fraction of PM is about 0.75 for the 2-stroke vehicles, with no detectable EC content (Figures 4.2B and 4.3C). Most of the PM2.5 from CNG-2S vehicles is likely aerosolized lubrication oil, which in CNG vehicles is injected 86  directly into the air intake and much of which may be emitted unburned (Faiz et al. 2004; Kojima et al. 2002). High PM2.5 emissions are thus a fundamental problem with many simple 2stroke engine designs, and not substantially reduced by switching to a clean fuel.  4.3.5. Assessing CNG for auto-rickshaws in Delhi Approximately 90% of Delhi’s auto-rickshaws – about 50,000 vehicles – have 4-stroke engines, and each vehicle travels about 55,000km each year (see Chapter 3). Using this study’s mean EFs, it is possible to calculate that switching the Delhi 4-stroke fleet from gasoline to CNG would reduce PM2.5 emissions by approximately 99 tonnes per year (50,000 vehicles " 55,000 km " [48-12 mg km-1]). This is probably a conservative (high) estimate, since the sampled PET-4S group (running on infrequently-used back-up gasoline systems) likely had higher emissions than the gasoline 4-strokes engines in auto-rickshaws prior to their conversion to CNG. In comparison, if the remaining 5,000 2-stroke auto-rickshaws in Delhi were phased out and replaced with CNG 4-stroke vehicles, PM would be reduced by an additional 96 tonnes per year (5,000 vehicles " 55,000 km " [362-12 mg km-1]). In other words, replacing the remaining 2stroke auto-rickshaws in Delhi would have nearly the same effect as switching from gasoline to CNG in 10 times as many 4-stroke vehicles. Our findings suggest that CNG fuel should be limited to use in 4-stroke engines to realize potential health benefits (low PM2.5 and THC, less CO than PET-4S) and climate benefits (lower fuel consumption, less GWC than CNG-2S). Although the 2-stroke engines examined in this study were all CNG-fueled, other studies suggest that 2-stroke engines fueled with gasoline or LPG (another popular ‘clean fuel’ in India) have similarly high PM2.5, CO and THC (ARAI 2007; Kojima et al. 2002). These studies (see EFs in Table C.8 in Appendix C) indicate that a switch from gasoline or LPG to CNG may bring a small PM emissions reduction (around a factor of 2), however emissions are still many times higher than the ‘acceptable’ levels represented by CNG 4-s vehicles. The average CNG-2S vehicle emitted nearly 3 orders of magnitude (700 times) more PM than the new CNG-4S test vehicle (V14). In places where a large proportion of three-wheelers still have 2-stroke engines, retrofit technologies can be used to reduce 2-stroke engine emissions. In the Philippines, for example, relatively low-cost kits were developed that enabled direct fuel injection while metering the 87  lubricating oil, reducing unburned fuel emissions and PM while improving fuel efficiency (Lorenz et al. 2005). Such kits require development of a custom retrofit design for each engine type targeted, but this would not be a barrier for 2-stroke three-wheelers in India since one manufacturer has made these vehicles and engines for decades. However, regulators should also consider following the example of Dhaka, Bangladesh, who banned use of 2-stroke vehicles completely (Begum et al. 2006). This study has shown the need for comprehensive test programs that develop EFs based on multiple vehicles: accurate health- and climate-relevant emission factors are essential inputs for vehicle emission inventories or models, so that appropriate policies can be implemented.  88  4.4. References ARAI. Draft report on emission factor development for Indian Vehicles. Automotive Research Association of India, Air Quality Monitoring Project – Indian Clean Air Programme: Pune, India, 2007: http://www.cpcb.nic.in/DRAFTREPORT-on-efdiv.pdf. Baidya, S.; Borken-Kleefeld, J. Atmospheric emissions from road transportation in India. Energy Policy. 2009, 37 (10), 3812-3822. Bajaj Auto Limited, Annual Reports; http://www.bajajauto.com/bajaj_investor_annual_report.asp. Begum, B. A.; Biswas, S. K.; Hopke, P. K. Impact of banning of two-stroke engines on airborne particulate matter concentrations in Dhaka, Bangladesh. J. Air Waste Manag. Assoc. 2006, 56 (1), 85-89. Bell, M.; Davis, D.; Cifuentes, L.; Krupnick, A.; Morgenstern, R.; Thurston, G. Ancillary human health benefits of improved air quality resulting from climate change mitigation. Environ. Health. 2008, 7 (1); DOI: 10.1186/1476-069X-7-41. Bond, T. C.; Streets, D. G.; Yarber, K. F.; Nelson, S. M.; Woo, J. H.; Klimont, Z. A technologybased global inventory of black and organic carbon emissions from combustion. J. Geophys. Res. 2004, 109 (D14203). Brook, R. D.; Rajagopalan, S.; Pope, C. A., III; Brook, J. R.; Bhatnagar, A.; Diez-Roux, A. V.; Holguin, F.; Hong, Y.; Luepker, R. V.; Mittleman, M. A.; Peters, A.; Siscovick, D.; Smith, S. C.; Whitsel, L.; Kaufman, J. D. Particulate matter air pollution and cardiovascular disease: An update to the scientific statement from the American Heart Association. Circulation. 2010, 121 (21), 2331-2378. EPCA. Review of existing cap on the number of three-wheelers in Delhi and its implications for pollution and congestion. Environment Pollution (Prevention & Control) Authority for the National Capital Region: Delhi, India, 2010, Report No. 34: http://www.indiaenvironmentportal.org.in/files/epca_0.doc. Faiz, A.; Gautam, S.; Gwilliam, K. M. Technical and policy options for reducing emissions from 2-stroke engine vehicles in Asia. Int. J. Veh. Des. 2004, 34 (1), 1-11. Forster, P., Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.; Haywood, J.; Lean, J.; Lowe, D.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Dorland, R. V. Changes in atmospheric constituents and in radiative forcing, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon, S. et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, 2007. Grieshop, A. P.; Reynolds, C. C. O.; Kandlikar, M.; Dowlatabadi, H. A black-carbon mitigation wedge. Nature Geo. 2009, 2 (8), 533-534. 89  HEI. Traffic-related air pollution: A critical review of the literature on emissions, exposure, and health effects. Health Effects Institute: Boston, MA, 2010, Special Report 17: http://pubs.healtheffects.org/view.php?id=334. Heywood, J. B. Internal Combustion Engine Fundamentals. McGraw-Hill: New York, NY, 1988, p.830. Hopke, P. K.; Cohen, D. D.; Begum, B. A.; Biswas, S. K.; Ni, B.; Pandit, G. G.; Santoso, M.; Chung, Y.; Davy, P.; Markwitz, A.; Waheed, S.; Siddique, N.; Santos, F. L.; Pabroa, P. C. B.; Seneviratne, M. C. S.; Wimolwattanapun, W.; Bunprapob, S.; Vuong, T. B.; Hien, P. D.; Markowicz, A. Urban air quality in the Asian region. Sci. Tot. Environ. 2008, 404 (1), 103-112. ICAT. International Center for Automotive Technology Website; www.icat.in. Iyer, N. V. Role of Three-Wheeled Vehicles in Urban Transportation in South Asia. Smart Urban Transport. 2003, 2 (2). Kandlikar, M. Air pollution at a hotspot location in Delhi: Detecting trends, seasonal cycles and oscillations. Atmos. Environ. 2007, 41 (28), 5934-5947. Kim Oanh, N. T.; Upadhyay, N.; Zhuang, Y. H.; Hao, Z. P.; Murthy, D. V. S.; Lestari, P.; Villarin, J. T.; Chengchua, K.; Co, H. X.; Dung, N. T.; Lindgren, E. S. Particulate air pollution in six Asian cities: Spatial and temporal distributions, and associated sources. Atmos. Environ. 2006, 40 (18), 3367-3380. Kojima, M.; Bacon, R. W.; Shah, J.; Mainkar, M. S.; Chaudhari, M. K.; Bhanot, B.; Iyer, N. V.; Smith, A.; Atkinson, W. D. Measurement of mass emissions from in-use two-stroke engine three-wheelers in South Asia. SAE Tech. Pap. 2002, 2002-01-1681. Kuhns, H. D.; Mazzoleni, C.; Moosmüller, H.; Nikolic, D.; Keislar, R. E.; Barber, P. W., Li, Z.; Etyemezian, V.; Watson, J. G. Remote sensing of PM, NO, CO and HC emission factors for onroad gasoline and diesel engine vehicles in Las Vegas, NV. Sci. Tot. Environ. 2004, 322, 123137. Lorenz, N.; Bauer, T.; Willson, B. Design of a direct injection retrofit kit for small two-stroke engines. SAE Tech. Pap. 2005, 2005-32-0095. Mazzi, E. A.; Dowlatabadi, H. Air quality impacts of climate mitigation: UK policy and passenger vehicle choice. Environ. Sci. Technol. 2007, 41 (2), 387-392. Mehta, R. History, politics and technology of CNG-diesel switch in Delhi, in Land Use, Transportation and the Environment. The Transport Asia Project: Pune, India. 2001: http://www.seas.harvard.edu/TransportAsia/Dec01Papers.htm. Narain, U.; Krupnick, A. The impact of Delhi's CNG program on air quality. Resources for the Future: Washington, DC, 2007: http://www.rff.org/Publications/Pages/PublicationDetails.aspx?PublicationID=17476. 90  Peng, R. D.; Bell, M. L.; Geyh, A. S.; Mcdermott, A.; Zeger, S. L.; Samet, J. M.; Dominici, F. Emergency admissions for cardiovascular and respiratory diseases and the chemical composition of fine particle air pollution. Environ. Health Perspect. 2009, 117 (6), 957-963. Pope, C. A., III; Dockery, D. W. Health effects of fine particulate air pollution: Lines that connect. J. Air Waste Manag. Assoc. 2006, 56 (6), 709-742. Reynolds, C. C. O.; Kandlikar, M. Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860-5865. Russell, L. M. Aerosol organic-mass-to-organic-carbon ratio measurements. Environ. Sci. Technol. 2003, 37 (13), 2982-2987. Singer, B. C.; Harley, R. A. A fuel-based motor vehicle emission inventory. J. Air Waste Manag. Assoc. 1996, 46 (6), 581-593. Smith, K. R.; Uma, R.; Kishore, V. V. N.; Zhang, J. F.; V Joshi, V.; Khalil, M. A. K. Greenhouse implications of household stoves: An analysis for India. Ann. Rev. Energy Environ. 2000, 25, 741-763. Subramanian, R.; Winijkul, E.; Bond, T. C.; Thiansathit, W.; Kim Oanh, N. T.; Paw-Armart, I.; Duleep, K. Climate-Relevant Properties of Diesel Particulate Emissions: Results from a Piggyback Study in Bangkok, Thailand. Environ. Sci. Technol. 2009, 43 (11), 4213–4218. West, J. J.; Fiore, A. M.; Horowitz, L. W.; Mauzerall, D. L. Global health benefits of mitigating ozone pollution with methane emission controls. PNAS. 2006, 103 (11), 3988-3993. Yeh, S. An empirical analysis on the adoption of alternative fuel vehicles: The case of natural gas. Energy Policy. 2007, 35 (11) 5865-5875.  91  5. Chapter 5: Fuels, technology and vehicle maintenance: Assessing strategies to reduce emissions from Indian auto-rickshaws  5.1. Introduction  Urban air pollution in developing countries is a serious health concern, and governments have been struggling to identify and control the sources of pollutants for decades (Mage et al. 1996). Over half of the global population now lives in cities, so the health and environmental impacts of urban air pollution are affecting more people than ever before. The problem is particularly acute in megacities (with over 10 million inhabitants). Recent multi-pollutant assessments of air quality (AQ) in megacities have shown that five South Asian megacities (Delhi, Dhaka, Karachi, Kolkata, and Mumbai) have exceptionally high ambient concentrations of particulate matter (PM), sulphur dioxide and nitrogen dioxide (Molina and Molina 2004; Gurjar et al. 2008). Emissions from transportation make up a substantial contribution to these and other air pollutant concentrations (HEI 2010). Knowledge gaps about the type and rate of emissions from motorized vehicles are acute, especially with respect to particulate matter (PM) air pollution (HEI 2010, Baidya and BorkenKleefeld 2009). This problem is exacerbated in less developed countries with resource constraints, due to the cost and complexity of collecting such data. We recently conducted a comprehensive study of particulate and gaseous emissions from in-use auto-rickshaws (threewheeled taxis) in Delhi, India (Reynolds et al. 2009; also see Chapter 4) as part of the Indian Auto-Rickshaw Project (IARP). Here we use data from IARP as well as other published autorickshaw emissions studies to model and assess the impact of a range of pollution control strategies aimed at reducing auto-rickshaw emissions. We focus on air pollutant emissions that are linked to adverse health impacts, but we also consider the climate impacts of the exhaust constituents.  92  Auto-rickshaws have relatively small, fuel-efficient engines that are similar to those used in most two-wheelers (i.e., motorcycles and mopeds) in developing countries. The IARP emissions dataset includes 42 chassis-dynamometer tests on 31 in-use vehicles with different engine and fuel types. The sample included vehicles with 4-stroke (4S) spark-ignited engines fueled with both gasoline/petrol (PET) and compressed natural gas (CNG), as well as 2-stroke (2S) engines fueled with CNG. To extend the IARP dataset, other gaseous and particulate emissions data from PET-4S and PET-2S auto-rickshaws – tested in an identical manner to IARP – are also incorporated in this paper (Kojima et al. 2002; ARAI 2007). Auto-rickshaw engines do not have some technologies now commonly used in automobiles, such as fuel injection or catalytic converters. Pollutant emission factors from other light-duty vehicles (LDVs) could be significantly different from auto-rickshaws depending on many factors, including (but not limited to) engine displacement and design, presence of exhaust aftertreatment devices, and level of maintenance. However, the modeling approach to policy evaluation introduced here can readily be applied to other empirical datasets. It may be especially effective if used with larger datasets, such as can be obtained from remote sensing or tunnel-based measurement campaigns that have emission factors for hundreds or thousands of vehicles (e.g., Ropkins et al. 2009). The emissions control strategies, (or policies), for the Indian auto-rickshaws evaluated in this paper include: •  New vehicle emission standards.  •  Phase out of 2-stroke (2S) engines and replacement with 4-stroke (4S) engines.  •  Switch from gasoline/petrol (PET) to CNG fuel10.  •  Scrapping of older vehicles in the fleet.  •  A range of inspection and maintenance (I/M) programs, based on idle and dynamometer testing of different pollutant species.  We focus on PM2.5 emissions since exposure to this pollutant is more strongly associated with adverse health effects than other traffic-related pollutants (Pope and Dockery 2006). In addition 10  To address Delhi’s air pollution problem, the Supreme Court mandated that all of its public transportation vehicles (auto-rickshaws, taxis and buses) must operate on CNG instead of gasoline (Narain and Krupnick 2007). CNG is also used in auto-rickshaws in Mumbai, Kolkata, and Dhaka.  93  to PM2.5 we examine carbon monoxide (CO), non-methane hydrocarbons (NMHC) and oxides of nitrogen (NOX), which are also important from a health perspective (HEI 2010). Increasingly, researchers have been investigating the relationship between health and climate impacts of combustion emissions; see for example the recent work on the health co-impacts of GHGmitigation efforts (Mazzi and Dowlatabadi 2007; Smith and Haigler 2008; Woodcock et al. 2009). In a similar manner, we quantify the climate impacts of the AQ policies examined in this study. In addition to Kyoto protocol emissions CO2 and methane (CH4), we evaluate the change in non-Kyoto species: CO, NMHC, and the elemental carbon (EC) and organic carbon (OC) components of PM (Forster et al. 2007). This paper is structured as follows: First, we present the vehicles considered in this analysis (Section 2.1) and their emission factors. The metrics used to represent health and climate impacts are described (Section 2.2). We provide details about the emission-reduction policies that we have assessed, and describe how they have been modeled in this study (Section 3). The impact of the policies on pollutant reduction and on climate forcing is discussed in Section 4. We conclude with a discussion of how these findings might be used to estimate actual health impacts, and the implications of this work for policy-makers in developing countries.  5.2. Methods: Quantifying health and climate-relevant emissions  5.2.1. Vehicles and fuels The baseline data for this study came from laboratory measurements of emissions and fuel consumption of Indian auto-rickshaws manufactured by Bajaj Auto Ltd. (Chapter 4 of this thesis; Kojima et al. 2002; ARAI 2007). Four different fuel/engine combinations were examined: PET4S, CNG-4S, PET-2S and CNG-2S. All vehicles were tested in an identical manner, on chassis dynamometers using the standardized Indian Drive Cycle (IDC). The IARP auto-rickshaws were randomly selected in order to be representative of the Delhi Auto-rickshaw fleet. The PET-4S vehicles in the IARP sample (MY 2000-2009) had dual-fuel systems that normally operated on CNG. All but the newest vehicles in the group had relatively high CO emissions and high fuel consumption, which is indicative of poor combustion. The PET-2S sample from Kojima et al. (2002) were MY 1993-1996 vehicles tested in early 2000s. Additional data for PET-2S and PET94  4S were included from a recent study by the Automotive Research Association of India (ARAI 2007). Differences in the emission factors from the different datasets have been discussed in Chapter 4 of this thesis. In this study we report ‘service-based’ emission factors, i.e., mass of pollutants emitted per distance traveled (g/km), to facilitate comparisons between policies. The emission factors (EF) for gaseous and particulate pollutants in each base group are shown in Table 5.1. Gaps in the data were identified: neither ARAI (2007) nor Kojima (2002) reported CH4 emissions or the OC and EC components of PM. Therefore, for the vehicles with PET-2S, it was assumed that 5% of the total hydrocarbon emissions were CH4, and it was assumed that the PM was composed of 75% OC and 5% EC (Alander et al. 2005; Bond et al. 2004). PET-4S emissions ratios CH4/NMHC, OC/PM and EC/PM were assumed to be the same from the ARAI data as from the IARP data.  Table 5.1. Fuel consumption and distance-based emissions factors for the base datasets; distribution in data is represented by median and inter-quartile range. PET-2S CNG-2S Sample size N = 46 N = 13 Fuel Consumption 3 (2.9, 4.2) 2.6 (2.2, 2.8) (kg/100km) CO2 (g/km) 49 (46, 52) 48 (46, 50) CO (g/km) 13 (7, 16) 1.1 (0.9, 2.1) CH4 (g/km) 0.42 (0.37, 0.78) 7.7 (6.9, 8.3) NMHC (g/km) 7.9 (7, 15) 1.6 (0, 2) NOX (g/km) 0.08 (0.04, 0.12) 0.06 (0.05, 0.07) OC (mg/km) 308 (200, 650) 142 (70, 220) EC (mg/km) 21 (10, 40) 0 (0, 0) a PM2.5 (mg/km) 410 (270, 870) 184 (110, 280) Data Source: A, B C a EC from CNG-2S below limit of detection. A = Kojima et al. 2002 B = ARAI 2007 C = Indian Auto-rickshaw Project (see Chapter 4)  PET-4S N = 13  CNG-4S N = 17  3.9 (3.3, 4.3)  2.1 (2, 2.2)  59 (47, 63) 27 (22, 36) 0.36 (0.3, 0.5) 2.3 (1, 9) 0.4 (0.1, 0.5) 19 (9, 32) 12 (9, 21) 30 (23, 62) B, C  55 (52, 57) 1.3 (0.6, 2.7) 1.1 (0.5, 1.5) 0.1 (0, 0.3) 0.4 (0.3, 0.6) 4.4 (0.6, 8.3) 1.7 (0.2, 3.1) 10 (1.1, 17.1) C  In India, new vehicles are required to meet emissions standards called the Bharat Stage (BS) Norms. Auto-rickshaws manufactured before April 2010 must meet BS II limits of 2.25 g/km CO and 1.25 g/km NOX+THC. BS II is approximately comparable to the European ‘Euro 2’ standard; the implementation of Indian standards lags Europe by about a decade. EFs from new auto-rickshaws, manufactured to meet BSII, can be taken as representing best-case emissions. 95  However, new vehicles are only required to meet BS standards over their first 30,000km, whereas most auto-rickshaws would be expected to travel further than that in their first year of use (auto-rickshaws are estimated to travel ~150 km per day on average; see Chapter 3). Emission factor data was available for a ‘new’ duel-fuel 4S auto-rickshaw with <5,000 km on its odometer, and EFs from this vehicle were assumed to represent lower bounds for the PET-4S and CNG-4S groups.  5.2.2. Calculating climate impacts CO2 and CH4 are the only species in this study that are included in the Kyoto Protocol, though neither are regulated by Indian air quality policies. CH4 emissions from vehicles are important to quantify (from a climate perspective) because CH4 has a 100-year global warming potential (GWP100) that is 25 times that of CO2 (Forster et al. 2007). When measuring emissions from vehicles fueled with CNG, which is >80% CH4, this is even more crucial. However other pollutants also have climate impacts, including the components of particulate matter: EC and OC (the latter reflects incoming radiation rather than absorbing it, so it has a cooling instead of a warming impact). GWP100 (mean, confidence interval) for the non-Kyoto CFAs considered in this study are CO: 1.9 (1.0 to 3.0); NMHC: 3.4 (1.7 to 6.8); OC: -35 (-9 to -83); EC: 455 (193 to 716) (Forster et al. 2007; Reynolds and Kandlikar 2008). The global warming commitment (GWC) of operating a vehicle can be expressed as the mass of CO2-equivalent emissions per km traveled. GWC is calculated by summing the impacts of each of the CFAs in the exhaust emissions, using the following equation: n  GWC" = EFCO2 + # EFi $ GWP",i  (Equation 1)  i=1  where EFi is the emission factor of exhaust constituent i, and GWP",i is the global warming  !  potential for exhaust constituent i over time-period ! (by convention, ! is usually 100 years). !  96  5.3. Modeling emission control strategies  The emission control policies under consideration for auto-rickshaws with spark-ignition engines are described below, and the criteria used to test the impact of each strategy on our data are summarized in Table 5.2. The policies were categorized into two main groups: (a) fuel/engine switching, and (b) I/M programs. (The policy to scrap old vehicles is the exception, and has been listed separately). We simulated the effect of policies in category (a) by comparing emissions across sample groups (e.g. PET-4S vs. PET-2S). In category (b) the I/M criteria were applied to each set of vehicles within a given fuel/engine group, and ‘failing’ high-emitters were removed from the group. The average emission factors for the post-policy group (containing only those vehicles that ‘passed’ the inspection in question) were then recalculated so they could be compared to the emissions from the base group.  97  Table 5.2. Model implementation criteria for each strategy to reduce auto-rickshaw emissions. Strategy  Group(s) affected Model implementation criteria  Fuel/Engine Switching: New Vehicle  PET-4S, CNG-4S  A single low-mileage 4S vehicle operating on CNG and PET (vehicle manufactured to meet Bharat Stage Norms; represents ‘lower bound’ for 4S emissions).  Replace 2S with 4S  All  Comparing base groups: PET-4S vs. PET-2S, and CNG-4S vs. CNG-2S; (assume no change in fuel type).  Switch to CNG  All  Comparing base groups: PET-4S vs. CNG-4S, and PET-2S vs. CNG-2S; (assume no change in engine type).  Inspection/Maintenance Programs: I/M: Idle Gaseous  PET-4S a  Vehicles must pass India’s idle emissions test, ‘Pollution Under Control’ (PUC Limits: 3.5% CO and 4500 THC).  I/M: Idle Visible PM PET-2S, CNG-2S b Vehicles must ‘pass’ a visible smoke/PM inspection.  I/M: Dyno Gaseous  I/M: Dyno PM  All  Vehicles must ‘pass’ a hypothetical inspection based on measurement of gaseous pollutants over the Indian Drive Cycle (IDC). Limits for 4S and 2S vehicles were set at 1.5 and 6 times the Indian Bharat Stage II Norms, respectively (BSII = CO < 2.25 g/km and THC+NOX < 2.0 g/km).  PET-2S, CNG-2S, PET-4S c  Vehicles must ‘pass’ a hypothetical inspection based on measurement of PM over IDC. Limits for 4S and 2S vehicles were set at 5 and 1 times the proposed BSIII Norms for PM from new diesel auto-rickshaws, respectively (BSIII = 0.05 g/km).  Other: Scrap Old Vehicles  PET-4S, CNG-4S d Vehicles MY 2001 and older scrapped.  a  IARP study vehicles all underwent PUC testing, however in the CNG-4S and CNG-2S groups, no vehicles ‘failed’ this test so it was only possible to model the effect of the policy for PET-4S. b IARP study vehicles were inspected for visible smoke (see Chapter 3), and vehicles from Kojima (2002) were tested with an opacity meter (emissions factors more than 1g/km PM were classified as emitting visible smoke). c CNG-4S vehicles all ‘passed’ so it was only possible to model the effect of the policy on that group. d Only MY 2007-2009 remained (no vehicles of MY 2002-2006 in the sample).  5.3.1. Replace 2-stroke engines 2-stroke engines are simple, reliable, and powerful for their weight. However, 2S engines produce far higher particulate matter (from lubricating oil that is mixed with the fuel in the combustion chamber) and THC (from unburned fuel) emissions than 4-stroke engines. Considerable effort has been expended to find ways of reducing emissions from 2S engines in Asia, because they are still used in many two-wheelers and three-wheelers (Faiz et al. 2004). Traditionally, 2S engines are preferred by many auto-rickshaw drivers because of their 98  simplicity and low cost, though some cities (such as Dhaka and Kolkata) have banned the use of 2S engines in three-wheeled commercial vehicles (Begum et al. 2006). Policy makers should be cognizant of equity considerations: an outright ban on 2S engines could adversely impact lowincome vehicle owners/drivers.  5.3.2. Switch from gasoline to alternative fuel The vast majority of light-duty vehicles are fueled with conventional liquid gasoline, but ‘clean’ alternative fuels such as CNG and liquefied petroleum gas (LPG) are increasingly being promoted in India as a way of reducing traffic-related emissions in cities. CNG fueling stations for motor vehicles have been already introduced in over 60 Indian cities including Delhi, Mumbai, Hyderabad, Pune, and Bangalore (PIB 2010), as well a number of cities in Pakistan11. Vehicles operators may be subjected to considerable costs and inconvenience if they are required by law to convert their vehicles, although government subsidies in India mean that CNG is about half the price of gasoline, which reduces fueling costs.  5.3.3. Inspection/maintenance program (Type I): Idle emissions testing Inspection and maintenance (I/M) programs are designed to identify high-emitting vehicles from the in-use fleet during regular, mandatory emission tests. The programs may be enforced by linking them with annual vehicle insurance or registration programs, or by fining drivers who are found to be not in compliance. In India, the I/M program is called ‘Pollution Under Control’ (PUC), and is based on measurement of CO and THC concentration in the undiluted exhaust while the engine is idling. For three-wheelers manufactured before March 2000, PUC limits are 4.5% CO and 9000ppm THC, while for vehicles manufactured after March 2000 the limits are 3.5% CO and 4500ppm THC. If the concentration of either pollutant exceeds the limit, the vehicle must be repaired by the owner and retested. PUC test data were available for the IARP test vehicles. We denote idle emissions testing based on gaseous species as ‘I/M: Idle Gaseous’ in our model.  11  A challenge with policies that encourage switching to CNG or LPG is the cost of fuel-supply infrastructure. CNG refueling stations are more costly than LPG because of the need to compress CNG to at least 20 atm for onboard vehicle storage. However investment in refueling infrastructure may be worthwhile if a natural gas supply is already in place in a city.  99  The Indian PUC program has been criticized as being ineffective for a number of reasons: idle emissions are poorly correlated with driving emissions, the instrumentation is simple and prone to malfunctioning, the tests are burdensome to drivers because they are overly frequent (biannual), and ‘test-and-repair’ garages are subject to widespread corruption (Hausker 2004; Rogers 2002). Another limitation of the PUC program is that it does not measure NOX or PM emissions. An alternative to measuring gaseous emissions at idle is to observe idling vehicles for visible smoke (i.e., PM) emissions. Despite the low-tech nature of this protocol, traffic authorities in Delhi have used this approach to identify high-emitters and to levy fines on drivers (EPCA 2004). We denote this approach ‘I/M: Idle Visible PM’. Visible smoke in 2-stroke engines is an indication that drivers are mixing too much lubricating oil with the fuel (Faiz et al. 2004).  5.3.4. Inspection/maintenance program (Type II): Chassis dynamometer testing I/M programs based on emissions measurements over a standardized drive cycle on a chassis dynamometer are better at identifying high-emitters than idle emission measurements, because the test more accurately represents real driving conditions (Hausker 2004). However, it takes substantial institutional and financial resources to successfully implement a dynamometer-based I/M program, so such policy approaches are more commonly found in developed countries. In India, auto manufacturers must demonstrate that new vehicles meet the Bharat Stage Norms using laboratory dynamometer testing with the Indian Drive Cycle (IDC). Since a dynamometerbased I/M program does not presently exist in India, here we implement hypothetical limits for auto-rickshaws by scaling up BSII norms by a factor of 1.5 and 6 for 4S and 2S vehicles, respectively. The choice of scaling factors was based on an iterative process, and aimed to identify a sufficient number of ‘high-emitters’ without ‘passing’ or ‘failing’ the entire group. This policy is denoted ‘I/M: dyno gaseous’. Measuring PM requires relatively complex, sensitive and expensive equipment, so few I/M programs measure PM from light-duty spark-ignition engines. In order to explore the potential for this approach to reduce emissions, a hypothetical I/M program was tested in our model that resembles the ‘dyno gaseous’ program described above, but has limits for PM2.5 emissions instead of gaseous pollutants (denoted ‘I/M: Dyno PM’). Since spark-ignited engines are not required to meet PM emissions limits, in our model we base the criteria on the BSIII limit for 100  diesel auto-rickshaws: 0.05 g/km. The model criteria for 4S spark-ignition engines were set at the BSIII limit (0.05 g/km). Because 2S spark-ignition engines are high-PM emitters, the model criteria for PET-2S and CNG-2S was scaled by a factor of 5 times the BSIII norm (0.25 g/km).  5.3.5. Scrapping old vehicles A commonly used regulatory tool for emissions control in developing countries is to simply require that all vehicles older than a certain model year be scrapped (or sold out of the jurisdiction). In Delhi, for example, auto-rickshaws and taxis older than 10 years are not allowed to operate. Implementation of that policy in Delhi was not without considerable difficulties, in part due to objections from drivers/owners and their unions (Narain and Bell 2005). Our model explores the impact of scrapping vehicles of 2001 model year or older in the 4S sample groups. There was a bi-modal age distribution in the IARP samples that represented the same distribution in the overall auto-rickshaw population in Delhi (with few vehicles of MY 2002-2006 inclusive). Therefore the ‘Scrap Old Vehicles’ criteria in our model resulted in only MY 2007-2009 vehicles in the post-policy group. Note that if vehicles are not actually scrapped, but rather sold to another jurisdiction for continued operation, the impact of high emissions is simply shifted – not eliminated.  5.4. Results and discussion  5.4.1. Impact of policies on emission factors Figure 5.1. shows how the modeled policies affect emissions of four pollutants: PM2.5, CO, NMHC, and NOX from all four fuel/engine combinations. We also conducted an uncertainty analysis; uncertainty is represented by median and interquartile range (25th percentile to 75th percentile) for the base datasets and for all of the policy scenarios (see Tables D.1 and D.2 in Appendix D, with emission factors for 2S and 4S vehicles respectively). Fig. D.1 (also in Appendix D) shows both AQ and climate emissions on one plot, and gives a graphical representation of this uncertainty.  101  Three main observations can be made from Fig. 5.1. First, phasing out 2S engines and replacing them with 4S results in the greatest reduction in PM emissions. Second, switching fuel from PET to CNG reduces air pollutants substantially (PM, CO and NMHC) in both 2S and 4S engines. However use of CNG in 2S engines is not recommended because the benefits accrued are far less than the potential reductions that CNG has to offer as a clean fuel. Third, I/M programs should be designed to treat vehicles with 2S engines differently from those with 4S engines. For 2S engines that have not yet been phased out of the fleet, a visible smoke inspection would be a simple and inexpensive way of cutting PM emissions, whereas for 4S engines a more sophisticated chassis dynamometer-based program would be required to reduce in-use emissions of PM, CO, and NMHC. These observations are discussed in more detail below, and compared to the other policies modeled in this study.  102  Figure 5.1. Base emission factor data and modeled policy-induced change in distance-based pollutant emissions factors. A. PET-2S vehicle group. B. CNG-2S vehicle group. C. PET-4S vehicle group. D. CNG-4S vehicle group. Note that the x-axes are not all at the same scale, so – to facilitate comparison across the panels – the number ‘10’ has been highlighted in red on each x-axis. Also, PM2.5 and NOX emission factors have been multiplied by a factor of 100 and 10, respectively, to make them possible to read on the same scale.  5.4.1.a Replace 2-stroke engines Regardless of what fuel is used, current technology 2S engines emit high levels of PM – in fact they emit a similar amount of PM per kg of fuel as diesel engines (~10 g/kg), albeit with 103  different OC/EC composition. Comparing Panels A and B with Panels C and D (Fig. 5.1), it is evident that PM emissions from PET-2S and CNG-2S engines (~500 mg/km and ~300 mg/km) are at least an order of magnitude higher than PM from PET-4S and CNG-4S (~25 mg/km and ~5 mg/km). THC emissions from 2S engines are also high because of ‘short-circuiting’ during the intake and exhaust processes: some unburned fuel inevitably gets swept into the exhaust, although modern 2S engine designs are improved in this respect. Possible disadvantages of switching from 2S to 4S include 3-8 times higher NOX emissions (because of higher combustions temperatures) and increased CO emissions (factor of 2) in the case of PET-2S to PET-4S. Increases in theses gaseous pollutants do not offset the potential AQ and health benefits of PM reduction, however. The fact that 2S engines are high-PM emitters is not new news (e.g. Faiz et al. 2004). Given the magnitude of emissions reduction that can be realized, however, scrapping 2S engines should be made a high priority.  5.4.1.b Switch from gasoline to alternative fuel Comparing emissions from gasoline-fueled vehicles (PET-4S and PET-2S) with CNG-fueled vehicles (CNG-4S and CNG-2S), it is apparent that using CNG can reduce PM and NMHC emissions substantially. In Delhi, all auto-rickshaws were converted to CNG fueling regardless of engine type, and at first glance this would seem to have been a smart policy. Converting PET2S to CNG-2S reduced PM emissions by about 40%, but CNG-2S PM emissions are still an order of magnitude more than if they were converted to less polluting 4S engines, with either PET or CNG fuel (Fig. 5.1.A). The cost and inconvenience of converting PET-2S to CNG-2S is significant, and resources might be better spent on replacing PET-2S engines with less polluting 4S engines, which could use either PET or CNG fuel and still realize a benefit.12 In 4S engines, our model finds that the switch from PET-4S to CNG-4S could reduce PM emissions by about 75%. In general, CNG fuel systems are also robust to tampering and less reliant on aftertreatment devices than liquid fuel systems. CNG-4S emissions appear to be remarkably stable, and relatively unaffected by the policies tested in the model (Fig. 5.1.D).  12  A cost-effectiveness analysis would be required to determine where resources would be best spent, but that is beyond the scope of this chapter.  104  5.4.1.c Inspection/maintenance programs The results suggest that I/M programs can be effective at reducing pollutant emissions from the in-use fleet, especially if appropriately targeted at sub-groups within the population. Although we recommend that 2S engines should be phased out in urban areas, during the phase-out period there are likely to be a significant number of 2-stroke vehicles remaining in the light-duty fleet. If an auto-rickshaw fleet still has vehicles with 2S engines, our model suggests that a simple inspection for visible smoke (‘I/M: Idle Visible PM’) can perform as well as (or better than) dynamometer testing (Figs. 5.1.A and 5.1.B). This approach, if widely adopted, could be an improvement on Delhi’s current ‘Pollution Under Control’ idle emissions testing system for reducing PM emissions, at least for 2S vehicles. Once vehicles with 2S engines are phased out, I/M programs must rely on instrumentation to measure pollutant EFs because 4S engines rarely emit visible smoke. Idle emission measurement of CO and THC (‘I/M: Idle Gaseous’) is used in India’s PUC program. Some PET-4S vehicles in our sample failed the PUC test, due to their excessive CO and NMHC emissions. The modeled average emissions from the remaining vehicles in the group had 30% lower CO, and there was a concurrent reduction in PM (by almost 50%) and NMHC (by 75%) – see Figure 5.1.C. Unfortunately idle emission testing was not consistently effective across a range of fuel/engine types, and it is possible that there was a spurious correlation between idle test failures and reduced PM in the PET-4S sample. PUC testing did not find that any of the vehicles in the CNG4S, PET-2S or CNG-2S samples had failing levels of emissions. We speculate that this was because (a) idle emissions are poorly correlated with driving emissions, and (b) the test protocol does not mandate measurement of NOX or PM/visible smoke emissions. Given these problems, it appears that it would be difficult to improve the effectiveness of the current Indian PUC program. An alternative type of I/M program, not currently used in India, is based on chassis dynamometer emission testing over a standardized drive cycle. This was simulated in our analysis (‘I/M: Dyno Gaseous’), and was found to be more robust than PUC for PET-4S (Fig. 5.1.C). According to our model, a dynamometer-based program such as this could reduce PM by 50% from the base PET-4S sample, while simultaneously reducing CO and NMHC to even lower levels than would be achieved with PUC testing. Emissions from the CNG-4S group were 105  not reduced when subjected to the same criteria, which further supports the hypothesis that CNG-4S is a robust low-emissions technology.  5.4.1.d Scrap old vehicles The modeled effect of scrapping older vehicles in PET-4S and CNG-4S groups was not significant. Compared to other policies examined in our study, we were not able to discern a significant impact of age-based vehicle retirement schemes in our model. However, the analysis was limited by the constrained range of vehicle ages in our sample population. Further investigation is needed (perhaps based on a larger dataset from remote sensing) since there is evidence that the brand new PET-4S produces far lower emissions than the fleet average.  5.4.2. Climate considerations To complement the AQ assessment, Figure 5.2 shows the global warming commitment (GWC) for the base data and each policy scenario. The GWC of each climate-forcing species in the exhaust (kg CO2-equivalent per kg fuel) shows the species’ relative importance from a climate perspective. Using this approach, it becomes clear that for PET vehicles, CO and NMHC emissions are almost as important as CO2. For the CNG fueled vehicles, CH4 is the most important non-CO2 exhaust constituent due to the high CH4 content of CNG. For CNG-4S group, the GWC of CH4 is only slightly less than that of CO2, while for CNG-2S the CH4 impact is three times that of CO2 due to higher unburned fuel emissions. Comparing the climate impact of 2S engine emissions, the CNG-fueled vehicles have over double the GWC of the gasoline-fueled group, which policy makers should be aware of if they are considering converting 2S vehicles to CNG.13 The components of particulate matter (EC and OC), have a small effect on GWC relative to the gaseous species.  13  CH4 is a highly stable and unreactive molecule, which makes it difficult to control with aftertreatment devices (such as oxidation catalysts or three-way catalysts).  106  Figure 5.2. Group average global warming commitment (g CO2-equivalent per km) for each fuel/engine type and policy. All climate-forcing agents are shown, including short-lived species (CO, NMHC, EC and OC) as well as the Kyoto protocol species (CO2 and CH4).  5.4.3. Integrating AQ and climate co-benefits The integrated results of this policy analysis are summarized in Table 5.3, which highlights examples of policies with potential for co-benefits (solid boxes) and policies that could cause tradeoffs (dashed boxes). Only one of the policies had substantial adverse climate co-impacts: when PET-2S were switched to CNG-2S there were high CH4 emissions. Though there were certainly a number of situations where co-benefits were identified, there were also several situations where policies had negligible effects on both AQ and climate-relevant emissions. For vehicles with 2S engines (red boxes), switching CNG-2S engines to CNG-4S engines had substantial AQ and climate co-benefits. In contrast, switching gasoline-fueled PET-2S engines to CNG fueling reduced air pollutant emissions but greatly increased climate-forcing emissions (due to high levels of unburned CH4 emissions). There were also examples for 4S engines (blue boxes): a chassis dynamometer-based I/M program could reduce both AQ and climate-forcing  107  emissions from the PET-4S group, whereas our model suggest that emissions from CNG-4S vehicles would be unaffected by this I/M program.  Table 5.3. Overview of AQ-climate policy assessment. Solid boxes give examples of policies that should be pursued, while dashed boxes give examples of less effective policies. PET-2S  CNG-2S  PET-4S  CNG-4S  Strategy  AQ  Clim.  AQ  Clim.  AQ  Clim.  AQ  Clim.  Replace 2S with 4S  !!  -  !!  !!  n/a  n/a  n/a  n/a  Switch to CNG  !!  XX  n/a  n/a  !  !  n/a  n/a  Scrap Old Vehicles  n/a  n/a  n/a  n/a  -  -  -  -  I/M: Idle Gaseous  n/a  n/a  n/a  n/a  !  -  n/a  n/a  I/M: Idle Visible PM  !!  !  !!  -  n/a  n/a  n/a  n/a  I/M: Dyno Gaseous  !!  !  -  -  !  !  -  -  n/a I/M: Dyno PM !! ! !! ! Policies marked ! or !! have a positive or very positive effect on tailpipe emissions Policies marked " or "" have a negative or very negative effect on tailpipe emissions ‘-’ indicates no significant effect observed; n/a: data not available  n/a  5.5. Ranking policy options  It is difficult to attribute health damages to climate emissions, since the adverse impacts of climate change take place over much larger spatial and temporal scales than urban air pollution. In contrast, the health impacts of exposure to urban air pollution (in particular PM2.5) are relatively well understood. Therefore, we rank a selection of the most promising policies in terms of their impact on PM2.5 emissions (Table 5.4). The results are presented as the mass of PM2.5 reduced annually from a fleet of 5,000 auto-rickshaws, due to implementation of each policy. The average annual distance traveled by each auto-rickshaw is assumed to be approximately 55,000 km (see Chapter 3). Figure D.2 (Appendix D) graphically depicts the relative PM reductions from the main policies for auto-rickshaws with 2S and 4S engines.  108  Table 5.4. Reduction in PM2.5 emissions (tonnes per 5,000 auto-rickshaws per annum) attributable to a given policy. Rank (most effective to least effective)  Policy  Policies targeted at 2-stroke vehicles 1 Switch fuel/engine: PET-2S to CNG-4S 2 Switch engine: PET-2S to PET-4S 3 Switch engine: CNG-2S to CNG-4S 4 Switch fuel: PET-2S to CNG-2S 5 PET-2S plus I/M Idle Visible PM 6 CNG-2S plus I/M Idle Visible PM Policies targeted at 4-stroke vehicles 7 Switch: PET-4S to CNG-4S 8 PET-4S plus I/M: Dyno Gaseous 9 PET-4S plus I/M: Idle Gaseous 10 Scrap old PET-4S 11 CNG-4S plus I/M: Idle Gaseous 12 CNG-4S plus I/M: Dyno Gaseous a Difference in PM2.5 emissions too small to characterize.  PM reduction (g/km)  PM reduction (tonnes/year)  0.49 0.47 0.26 0.22 0.21 0.13  134 130 73 61 57 35  0.015 0.014 0.008 0.003 0.000 0.000  4.0 3.8 2.3 0.7 a a  The ranking in Table 5.4 is useful as a metric of policy effectiveness, but there are some important caveats that need to be highlighted. First of all, effectiveness is narrowly defined here as the impact of a policy on PM2.5 emissions. Implementation costs, resource constraints, and socio-economic impacts experienced by auto-rickshaw drivers are not in the scope of this study, but should be included in a full assessment of options being considered for a city. Furthermore, the impact of any given policy is dependent on the number of vehicles it affects in the fleet in question. For example, in Delhi, there are presently approximately 50,000 CNG-4S autorickshaws and approximately 5,000 CNG-2S auto-rickshaws. Switching 5,000 CNG-2S autorickshaws to CNG-4S engines would result in around 73 tonnes PM2.5 avoided per annum, and there appear to be no comparable options to further reduce PM2.5 emissions from the 50,000 CNG-4S auto-rickshaws. Calculation of the health benefit of the policies (in terms of avoided mortality or morbidity for a given city) is beyond the scope of this paper. However the emission factors presented in the paper would be critical inputs for such a calculation. Health effects assessments are an important component of cost-benefit analyses. One way of calculating health benefits of a policy is to follow the methodology of Takeuchi et al. (2007). First, the reduction in ambient concentrations 109  of PM2.5 in a given city is estimated, using data about the fleet composition and atmospheric modeling. Second, results from air pollution epidemiology in developing countries (e.g., Wong 2008) can be used to estimate the proportional reduction in mortality (all-cause, cardiovascular, and/or respiratory) due to short-term PM2.5 exposure. Finally, overall hospital mortality data is needed for the jurisdiction in order to estimate the numbers of deaths avoided that can be attributable to the policy. Uncertainty in each step should be quantified and propagated through the assessment.  5.6. Conclusions  It may be tempting to look for a single optimal policy for reducing the AQ and climate impacts of auto-rickshaw emissions, but the picture is more complex. In some cases there are adverse coimpacts, such as increased GWC if 2-stroke engines are switched from gasoline to CNG, a policy that improves AQ. Therefore we suggest that the most promising way to ensure that policy actions yield co-benefits is to implement a package of several policies. Consider a typical Indian auto-rickshaw fleet with both PET-2S and PET-4S vehicles, in a city where the authorities are considering switching to ‘clean’ fuel to reduce air pollution. In general terms, the findings of this study would lead to the following recommendations: (a) phase out use of PET-2S engines in auto-rickshaws, and do not convert them to CNG; (b) introduce inspections for visible smoke for in-use PET-2S [if (a) is not followed or while there are 2S vehicles remain in the fleet] and ensure ‘smoking’ vehicles are repaired; (c) improve the I/M program, with the aim of transitioning to dynamometer-based testing at centralized, test-only centres; and, (d) consider switching from PET-4S to CNG-4S. The marginal benefits of implementing the latter two options are significantly less than the benefits of targeting 2-stroke engines. Nonetheless, in developing countries, switching to CNG (for 4S vehicles) can be an effective and robust means of reducing tailpipe pollutant emissions from light-duty vehicles, especially if the institutional or economic barriers to establishing a reformed I/M program are too great.  110  5.7. References  Alander, T.; Antikainen, E.; Raunemaa, T.; Elonen, E.; Rautiola, A.; Torkkell, K. Particle emissions from a small two-stroke engine: Effects of fuel, lubricating oil, and exhaust aftertreatment on particle characteristics. Aerosol. Sci. Tech. 2005, 39 (2), 151-161. ARAI. Draft report on emission factor development for Indian Vehicles. Automotive Research Association of India, Air Quality Monitoring Project – Indian Clean Air Programme: Pune, India, 2007: http://www.cpcb.nic.in/DRAFTREPORT-on-efdiv.pdf. Baidya, S.; Borken-Kleefeld, J. Atmospheric emissions from road transportation in India. Energ. Policy 2009, 37 (10), 3812-3822. Begum, B. A.; Biswas, S. K.; Hopke, P. K. Impact of banning of two-stroke engines on airborne particulate matter concentrations in Dhaka, Bangladesh. J. Air Waste Manag. Assoc. 2006, 56 (1), 85-89. Bond, T. C.; Streets, D. G.; Yarber, K. F.; Nelson, S. M.; Woo, J. H.; Klimont, Z. A technology-based global inventory of black and organic carbon emissions from combustion. J. Geophys. Res. 2004, 109 (D14203). EPCA. Report on the increase in the number of three-wheelers in Delhi. Report No. 9. Environment Pollution (Prevention & Control) Authority for the National Capital Region: Delhi, India, 2004. Faiz, A.; Gautam, S.; Gwilliam, K. M. Technical and policy options for reducing emissions from 2-stroke engine vehicles in Asia. Int. J. Veh. Des. 2004, 34 (1), 1-11. Forster, P., Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.; Haywood, J.; Lean, J.; Lowe, D.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Dorland, R. V. Changes in atmospheric constituents and in radiative forcing, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon, S. et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, 2007. Gurjar, B. R.; van Aardenne, J. A.; Lelieveld, J.; Mohan, M. Emission estimates and trends (1990-2000) for megacity Delhi and implications. Atmos. Environ. 2004, 38 (33), 5663-5681. Hausker, K. Vehicle inspection and maintenance programs: International experience and best practices. US Agency for International Development, Office of Energy and Information Technology: Washington, DC, 2004: http://pdf.usaid.gov/pdf_docs/PNADB317.pdf. HEI. Traffic-Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Special Report 17; Health Effects Institute: Boston, MA, 2010. Kojima, M.; Bacon, R. W.; Shah, J.; Mainkar, M. S.; Chaudhari, M. K.; Bhanot, B.; Iyer, N. V.; Smith, A.; Atkinson, W. D. Measurement of mass emissions from in-use 111  two-stroke engine three-wheelers in South Asia. SAE Tech. Pap. 2002, 2002-011681. Mage, D.; Ozolins, G.; Peterson, P.; Webster, A.; Orthofer, R.; Vandeweerd, V.; Gwynne, M. Urban air pollution in megacities of the world. Atmos. Environ. 1996, 30 (5), 681-686. Mazzi, E. A.; Dowlatabadi, H. Air quality impacts of climate mitigation: UK policy and passenger vehicle choice. Environ. Sci. Technol. 2007, 41 (2), 387-392. Molina, M. J.; Molina, L. T. Megacities and atmospheric pollution. J. Air Waste Manag. Assoc. 2004, 54 (6), 644-680. Narain, U.; Bell, R. G. Who Changed Delhi's Air? The Roles of the Court and the Executive in Environmental Policymaking. RFF DP 05-48. Resources for the Future: Washington, DC, 2005. Narain, U.; Krupnick, A. The impact of Delhi's CNG program on air quality. RFF DP 07-06. Resources for the Future: Washington, DC, 2007. PIB. Effects of Introduction of Gaseous Fuels for Public Transport. Press Information Bureau, Government of India, 2010: http://pib.nic.in/release/release.asp?relid=65189. Pope, C. A., III; Dockery, D. W. Health effects of fine particulate air pollution: Lines that connect. J. Air Waste Manag. Assoc. 2006, 56 (6), 709-742. Reynolds, C. C. O.; Kandlikar, M. Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860-5865. Reynolds, C. C. O.; Grieshop, A. P.; Kandlikar, M. Testing Emissions from auto-rickshaws to inform better air pollution control policies, SIM-air Working Paper Series 28, 2009: http://www.urbanemissions.info/sim-series-28.html. Rogers, J. Assessment of the pollution under control program in India and recommendations for improvement. World Bank, South Asia Urban Air Quality Management Program: Washington, DC, 2002: http://siteresources.worldbank.org/PAKISTANEXTN/Resources/UrbanAir/MainReport.pdf. Ropkins, K.; Beebe, J.; Li, H.; Daham, B.; Tate, J.; Bell, M.; Andrews, G. Real-world vehicle exhaust emissions monitoring: Review and critical discussion. Crit. Rev. Environ. Sci. Technol. 2009, 39 (2), 79-152. Smith, K. R.; Haigler, E. Co-benefits of climate mitigation and health protection in energy systems: Scoping methods. Ann. Rev. Pub. Health. 2008, 29, 11-25. Takeuchi, A.; Cropper, M.; Bento, A. The impact of policies to control motor vehicle emissions in Mumbai, India. J. Reg. Sci. 2007, 47 (1), 27-46.  112  Wong, C.; Vichit-Vadakan, N.; Kan, H.; Qian, Z. Public Health and Air Pollution in Asia (PAPA): A Multicity Study of Short-Term Effects of Air Pollution on Mortality. Environ. Health Perspect. 2008, 116 (9), 1195-202. Woodcock, J.; Edwards, P.; Tonne, C.; Armstrong, B. G.; Ashiru, O.; Banister, D.; Beevers, S.; Chalabi, Z.; Chowdhury, Z.; Cohen, A.; Franco, O. H.; Haines, A.; Hickman, R.; Lindsay, G.; Mittal, I.; Mohan, D.; Tiwari, G.; Woodward, A.; Roberts, I. Public health benefits of strategies to reduce greenhouse-gas emissions: urban land transport. Lancet. 2009, 374 (9705), 1930-43.  113  6. Chapter 6: General Conclusions  6.1. Summary of thesis objectives  The overall objective of this dissertation was to quantify the effectiveness of emission control policies for in-use vehicles in India. The research described here was motivated by the urgent need for reliable data about transportation emissions in developing countries, given the extremely poor air quality in many cities (Molina and Molina 2004). In India, urban transport has reached a critical juncture (Pucher et al. 2005): there are increasing numbers of new vehicles on the road, but old, high emitting vehicles are not being removed, and infrastructure development has lagged far behind. Meanwhile policies to improve urban air quality have rarely been preceded by analyses of their potential impacts, neither have they been followed up with studies to gauge (and learn from) their successes and costs. One of the primary reasons for these knowledge gaps is that it is challenging to collect adequate data about vehicle activity and emissions, because there is huge variation in the types and condition of on-road vehicles, and policy analysts face the hurdle of limited resources (both financial and institutional). This doctoral thesis represents an attempt to tackle a number of data-collection issues head-on, by conducting a series of related empirical research projects in India. I have focused on understanding the impacts of the large-scale adoption of compressed natural gas (CNG) as an alternative to diesel and gasoline in New Delhi. More specifically, I have conducted an in-depth study of the activity patterns and emissions of three-wheeled auto-rickshaws. In Chapter 2, my objective was to quantify the climate impacts of switching to CNG for public transportation vehicles (auto-rickshaws, taxis and buses). In Chapter 3, I conducted a survey of 350 autorickshaw drivers to learn about their driving patterns, fuel consumption and CO2 emissions, as well as to better understand the determinants of high pollutant emissions (for example engine type, vehicle age, maintenance practices). The aim of the research described in Chapter 4 was to measure exhaust emissions from a representative sample of real-world auto-rickshaws, using both standardized and novel instrumentation in a controlled laboratory setting. The objective of Chapter 5 was to use the accumulated knowledge from the previous three studies to conduct a quantitative policy analysis: it examined a range of strategies used in India to reduce emissions 114  from auto-rickshaws. These four studies presented here form a cohesive, policy-relevant body of work that contributes to the understanding of how auto-rickshaws contribute to air pollution in cities in the developing world. Since the research conducted for this thesis primarily involved the three-wheeled auto-rickshaws that are ubiquitous in Asian cities, the overarching project has been dubbed The Indian AutoRickshaw Project (IARP). The remainder of this concluding chapter is structured as follows: First I discuss the steps that I took to make IARP a reality, since I have been one of the principal investigators on this project from its inception (Section 6.2). It is hoped that this information will be useful to future researchers who wish to conduct productive fieldwork related to transportation emissions in India or other developing countries. Next I give a chapter-by-chapter synthesis of the main research findings presented in this dissertation (Section 6.3). The work described in this thesis is part of a larger suite of projects that will answer a wide-ranging set of questions about emissions from vehicles in developing countries. Finally, I have described the main limitations of the body of work described in this dissertation (Section 6.4), and the implications of the research for policy makers (Section 6.5).  6.2. Fieldwork: The Indian Auto-Rickshaw Project (IARP)  The fieldwork conducted for this dissertation consisted of three trips to Delhi, in 2007, 2008 and 2009, each of around two months duration. The first research trip (December-January 2007) was largely an information gathering exercise, during which IARP was initially conceived and developed. The trip began with a workshop on ‘Transport, Health, Environment and Equity in Indian Cities’, and was hosted by the Transport Research and Injury Prevention Programme (TRIPP) at the Indian Institute of Technology, Delhi (IITD). Over 60 participants attended the workshop, including: academic researchers, senior government decision makers, and representatives from the industry and NGOs. All had expertise in topics related to urban transportation, such as emissions/air pollution, road safety and access for the urban poor and non-motorized modes. The aim of the workshop was to identify opportunities for urban transportation policy in India to address these issues in an integrated fashion, which offered me the opportunity to learn about how I might contribute to addressing critical knowledge gaps. Following the workshop, I developed the framework of IARP and arranged one-to-one, in-depth 115  meetings with over 20 people from academia, industry and the government. These meetings were extremely enlightening, dispelling many misconceptions and highlighting a number of important opportunities. I regard this exploratory process of open-minded information gathering as a crucial stage that contributed to the success of IARP. During this time I began designing the survey of auto-rickshaw drivers and the developing a methodology for laboratory measurement of emissions from ‘real-world’ vehicles on a chassis dynamometer at an Indian facility. My network of transportation experts in India were invaluable in providing critical feedback on these research proposals. The second research trip to Delhi was in November-December 2008, and the purpose was to conduct the survey of auto-rickshaw drivers. An essential element of this trip was working with an NGO called the Institute for Democracy and Sustainability (IDS). IDS does advocacy work for often-ignored road-users in Delhi, including cyclists, pedal-rickshaw drivers, and autorickshaw drivers, and its staff have worked on research projects (including other surveys) with faculty and students from the Tranport Research and Injury Prevention Program at IITD. I forged a strong relationship with Dr. Rajendra Ravi, the head of IDS, and we subsequently engaged him to help us recruit test subjects for the laboratory study described in Chapter 4. The survey required hiring and training five Hindi-speaking research assistants to conduct interviews with the drivers, and an additional person to conduct data entry. I worked closely with Dr. Ravi to recruit the staff, and we arranged an intensive training workshop, (including pilot data collection) prior to finalizing the survey instrument. Data collection was conducted in December 2008, and data entry of 350 completed surveys was completed by mid-January 2009. A further 31 surveys were completed by drivers of the auto-rickshaws that were brought to the engine testing facility. During that second trip, I also began doing the groundwork for the final IARP study, which would entail the laboratory-based chassis-dynamometer testing of a sample of in-use Indian auto-rickshaws. This involved liaising with my contacts at the International Centre for Automotive Technology (ICAT) in Manesar, near New Delhi, which is a fully-equipped vehicle testing and regulatory approval facility that was developed in a partnership between the Indian Government and the Indian auto manufacturing industry (Figure 6.1). At that early stage my goals were: to determine the scope of the measurements that could be performed at the facility, 116  to identify a window of opportunity to conduct the study, and to begin negotiating the terms of the contractual agreement between UBC and ICAT. I was able to finalize the details of the research project in the subsequent period, while I was based in Canada. However the value of first building good relations with future research collaborators needs to be emphasized here. It encouraged both parties to work harder to resolve challenges during the research preparation stage, despite the inconvenience of being based on different continents with time zones about 13 hours apart. I returned to India to conduct the work in September-October 2009. Andy Grieshop, a postdoctoral research fellow, joined me for this part of the fieldwork. Andy worked closely with me to develop the emissions measurement methodology, and designed and built some specialized research tools to collect particulate matter samples. We conducted a comprehensive set of measurements of both gaseous and particulate exhaust constituents. In additional to standard drive-cycle average emission factors, we also obtained real-time emissions measurements (at 1 Hz sampling rate) for the entire drive-cycle. For particulate matter, we collected samples on Teflon filters so that we could estimate mass PM emissions, ‘quartzbehind-Teflon’ samples to correct for positive artifacts, and a parallel set of quartz filters for speciation of the organic component of the PM. In addition, Dr. Steve Rogak (Mechanical Engineering, UBC) lent me a thermophoretic sampler to collect particle samples on 3mm diameter amorphous carbon grids, for subsequent transmission electron microscopy. Several spin-off projects have arisen as a direct consequence of these additional data-collection exercises, which address some limitations in the core IARP studies that make up my dissertation.  117  Figure 6.1. Indian Auto-rickshaws being tested at the International Center for Automotive Technology in Manesar, near New Delhi.  6.3. Synthesis of thesis findings  In Chapter 2 (Reynolds and Kandlikar 2008), I examined the climatic impacts of New Delhi’s policy to switch all on-road public transportation vehicles from conventional fuels to CNG. Because natural gas is composed mostly of methane (CH4, a potent greenhouse gas), my hypothesis was that climate impacts would increase following conversion to CNG. Particulate matter emissions are also important, because they are composed of elemental carbon (EC, which absorbs radiation because of its dark colour, and warms the atmosphere), organic carbon (OC, reflects radiation), and sulfate (also reflective). The GWP for the constituents of particulate matter are complex and uncertain, so estimates (with confidence interval to represent uncertainty) were calculated from radiative forcing and atmospheric lifetime data provided by the IPCC (Forster et al. 2007). A sensitivity analysis was performed to explicitly address uncertainty in GWP and emission factors. The study found that emissions of CH4 and PM from diesel buses were both critical contributors to the change in global warming commitment (i.e., CO2-equivalent emissions, CO2-eq). Overall, the policy to switch to CNG resulted in a 30% increase in CO2-eq from CH4. However, when PM is taken into account in our model, most of the adverse climate impact was offset by a reduction in EC from the converted bus engines. The net effect of the switch was estimated to be a 10% reduction in CO2-eq. There is significant potential for reducing the climate impacts of fleet conversions to CNG in developing countries, 118  if engine technologies are used that minimize CH4 emissions. For Chapter 3, I designed and conducted a structured survey of auto-rickshaw drivers in Delhi. By asking drivers about their auto-rickshaws, about how far they traveled each day, and about their fueling costs and maintenance practices, it was possible to assemble a detailed quantitative picture of the fleet. Analysis of the survey data yielded policy-relevant information about the characteristics of the fleet such as engine types, vehicle activity, fuel consumption and CO2 emissions. Delhi auto-rickshaws were estimated to travel approximately 150 km per 24-hour period. Vehicles with 4-stroke engines made up 90% of the fleet, and had about 20% better fuel consumption (and lower CO2 emissions) than auto-rickshaws with 2-stroke engines. Each vehicle was inspected for visible smoke at engine start-up. There was found to be good correlation between those identified as ‘high-PM emitters’ through the visual inspection process, with those identified as having high PM through standard laboratory measurements. Autorickshaws with 2-stroke engines were much more likely to be categorized as high-PM emitters than those with 4-stroke engines (odds ratio increased by a factor of 2.6, p<0.01). Within the group of 4-stroke vehicles, age was a highly significant predictor of high-emitters (odds ratio increases by a factor of 0.18 for each additional year of age). The results of this study suggested that a simple observational test for visible smoke could be used as part of an inspection and maintenance (I/M) program to rapidly identify potential ‘super-emitters’ in a fleet. Auto-rickshaws are a ubiquitous element of urban passenger transportations systems in the developing world, yet there is sparse published information about their emissions. Chapter 4 in this dissertation describes a study that has addressed this knowledge gap, with a focus on 2stroke and 4-stroke spark-ignition engines fueled with CNG. A total of 41 chassis dynamometer emission tests were completed in a controlled laboratory setting. 30 test vehicles were recruited from the in-use fleet in Delhi and nearby Gurgaon, of which 11 were dual-fuel and operable on either gasoline or CNG. Despite high inter-vehicle variability, fuel-based emission factors were determined for gaseous pollutants (CO2, CH4, NOX, THC, and CO) and fine particulate matter (PM2.5). PM emission factors were of special interest because of their potential for adverse health and climate co-impacts. Auto-rickshaws with CNG 2-stroke engines emitted almost thirty times more PM2.5 (mean: 14.2 g kg-1, 95% confidence interval: 6.2-26.7) than those with CNG 4-stroke engines (0.5 g kg-1 [0.3-0.9]) and twelve times higher than for gasoline 4-stroke engines (1.2 g 119  kg-1 [0.8-1.7]). The global warming commitment of emissions from CNG 2-stroke engines was more than twice that of 4-stroke engines operating on either CNG or gasoline, due to high CH4 emissions. These measurements confirm that 2-stroke engines – even if fueled with ‘clean’ natural gas – emit high levels of particulate matter and methane. More generally, these data will be of value to researchers and policy-makers who plan to evaluate the impacts of CNG fueling in light-duty vehicles. Comprehensive measurements such as these should drive policy interventions, rather than assumptions about the impacts of clean fuels. Chapter 5 integrates the findings and methodologies of the prior chapters in a quantitative policy analysis. A range of emissions-reduction policies were evaluated using data from the study described in Chapter 4. The data on CNG-fueled auto-rickshaws was supplemented with additional emissions data from published studies of gasoline auto-rickshaw emissions. Emissions-control policies modeled include phasing out 2-stroke engines, switching to CNG fuel, scrapping older vehicles and four different types of I/M programs. Our results emphasize the urgency of phasing out the use of 2-stroke engines in urban areas, due to their high PM emissions. I/M programs are expensive and notoriously difficult to implement successfully in developing countries. However, when 2-stroke engines are still present in a fleet of autorickshaws, a simple I/M program based on inspection for visible smoke could reduce PM by almost 50%. (This finding confirms the observations made in Chapter 3.) For gasoline-fueled 4stroke engines, the model finds that a chassis dynamometer-based I/M program would be required to achieve the same proportional reduction in PM. This study confirms that robustly ‘clean’ technologies, such as CNG-fueled 4-stroke engines, can be used to reduce the need for I/M programs.  6.4. Limitations of the dissertation  In the studies described in this dissertation, I have focused on quantifying tailpipe emissions, not their impact on ambient concentrations of air pollution. The scope of this work therefore does not include an evaluation of the change in population exposure to air pollution, nor does it include estimation of health effects (in terms of mortality or morbidity, for example). In Chapter 5, however, I have outlined the steps that could be taken to translate change in emissions to a modeled change in ambient air quality. Using emerging epidemiological evidence about the 120  health impact of ambient particulate matter in Asian cities (Wong et al. 2008), it would be possible to estimate mortality attributable to changes in PM concentration. Few studies have attempted to evaluate the health impacts of transportation policies for developing countries, and this work is urgently needed if the true health cost of policies is to be calculated. The only example that I am aware of is a study by Takeuchi et al. (2007), which examined the health effects of transportation policies such as CNG switching in Mumbai. Future work could examine health impacts of changes in PM concentration in Delhi using our new knowledge about the emission factors from auto-rickshaw engines fueled with CNG. It is also possible that a healthimpacts study could be made even richer by using either the ‘intake-fraction’ approach (e.g., Bennet et al. 2002, Marshall et al. 2003), or to account for spatial variation in population and emissions that results in higher exposures to traffic related emissions near roadways (e.g., Gouge et al. 2010). In this thesis I have examined emissions and activity of auto-rickshaws (apart from Chapter 2, where the climate impacts of buses and taxis are also considered). Other vehicle types are undoubtedly also important sources of emissions, and it should not be inferred from this body of work that auto-rickshaws are the only – or even the primary – cause for concern. A limitation of measuring and modeling particulate matter mass emission factors is that it gives no information about the morphology, size distribution, or composition of the particles. In Chapter 4, I have provided information about the OC and EC components of the PM samples from auto-rickshaws, but additional work is needed to understand the details of PM morphology and size distribution (through transmission electron microscopy), and composition (through gas chromatography/mass spectrometry analysis). Other issues with the laboratory measurement of auto-rickshaw emissions are the assumptions that (a) the Indian Drive Cycle (IDC) is representative of real world emissions, and (b) that pollutants are emitted at a uniform rate, and independently of engine speed and load – neither of which are correct. In most instances, the errors introduced by these assumptions are small and of little concern because standardized drive cycles such as the IDC have been designed to reasonably approximate real-world emissions, and they suffice to compare emissions from one vehicle type to another. However, it is valuable to be able to ascertain approximately how 121  different the laboratory emissions are from reality. In addition, it can be important to know what engine conditions result in the highest emission rates: this information would enable us to more accurately model spikes in emissions during traffic activity, and hence model exposure to those pollutants. During the laboratory testing we were also able to obtain real-time emission measurements for both gaseous (CO2, CO, THC, and NOX) and PM2.5 emissions.14 These data – together with GPS activity data from an auto-rickshaw in Delhi – will enable future work to address this limitation by comparing the IDC to real-world driving by using a model that ‘bins’ second-by-second emissions by vehicle speed. A final limitation of this body of work is that the policy actions (or ‘strategies’) under consideration have not been evaluated from the perspective of their social and economic impacts. Rather, in this thesis my focus has been on measuring and modeling the impacts of the actions that result when a policy is implemented. Therefore although socio-economic studies were beyond the scope of this thesis, their importance is acknowledged here. Follow-on work is needed to provide a socio-economic assessment of the impacts of the CNG switch in Delhi from both the drivers’ perspective and in terms of institutional/implementation costs. Costs of purchasing/renting, fueling and maintaining CNG auto-rickshaws should be examined, and compared with the economic benefits/costs at the institutional level related to implementation as well as health and climate impacts (e.g. Kandlikar et al. in press).  6.5. Policy implications of the research  A challenge for researchers and policy-makers working on the impacts of transportation in the developing world has been the dearth of reliable emissions and activity data for ‘real-world’ vehicles. It is costly and resource-intensive to collect primary data (through conducting laboratory emission measurements and collecting activity data through surveys), and the resources to perform such studies in developing countries are limited. The research described in this thesis uses an integrated, interdisciplinary approach to fill knowledge gaps that have been identified, and the aim was to contribute towards informed, efficient policy decisions. There are limitations to the extent of this work, because of the significant costs and large scale of the 14  A DustTrak™ (TSI Instruments Model 8520) with a 2.5 micron cut-off inlet nozzle was used to measure real-time concentration of PM2.5, and was calibrated against filter measurements for each test.  122  measurements needed to quantify the emissions from such a large and dynamic population of vehicles. However, in this thesis I have presented some important (and sometimes counterintuitive) findings that will contribute to the understanding of urban motor vehicle emissions and their air-quality and climate co-impacts. Switching to a ‘clean’ alternative fuel does not result in a guaranteed reduction of both healthcritical pollutants and climate-forcing exhaust constituents. In certain cases, there may even be a health disbenefit if the conversion to a clean fuel causes a delay in upgrading to a newer engine technology. An example of the strong connection between air quality and climate impacts is given in Chapter 2: when heavy-duty diesel engines for buses were switched to CNG fuel there was a reduction in PM emissions, but a concurrent increase in CH4 emissions. Although there was a net climate benefit, (because of the near-elimination of warming diesel PM emissions), there could have been a significantly larger climate benefit if the converted CNG engines emitted less CH4. A second important example was found in the Delhi auto-rickshaw fleet. The opportunity cost of converting the existing gasoline-fueled 2-stroke engine to CNG was that they were not instead phased out and replaced with lower-emitting 4-stroke engines. The converted 2stroke engines still emitted high PM emissions, and converting them to CNG resulted in an order of magnitude increase in climate-forcing CH4 emissions. 2-stroke engines have long been recognized as having particularly high PM emissions (Faiz et al. 2004), but the policy-makers who mandated the CNG switch in Delhi did not take this into account. As a result, the drivers who were required to switch their 2-stroke vehicles to CNG fuel may have suffered significant inconvenience and cost without a significant emissions benefit. Another issue of relevance to climate policy is that PM emissions from diesel engines in developing countries can have very substantial climate impacts – this has been incorporated in the analysis conducted in Chapter 2. PM from diesel vehicles has a high proportion of elemental carbon, which absorbs radiation (due to its dark colour) and warms the atmosphere. Why should policy makers care about the climate impacts of air quality policy? There may be opportunities to leverage funding through the UNFCCC Clean Development Mechanism (CDM) or other joint implementation programs if air-quality and climate co-benefits can be identified. At present it is administratively challenging to get transportation projects recognized by the CDM. This is because of difficulties with establishing ‘baseline’ emissions and proving that a given policy 123  provides ‘additionality’ (i.e., a reduction in CO2-eq emissions beyond what would have occurred during business as usual). From a different perspective, it is increasingly being recognized in the developing world that climate change is a global problem that will require global action, so if a solution offers improvements in both air quality and climate co-benefits, then it should be chosen over a similarly expensive program that offers only air quality benefits (or that has climate disbenefits). It may be tempting to look for a single optimal policy to reduce traffic related emissions and their health and climate co-impacts, but the preceding discussion reveals that there are multiple possible approaches with complex interactions and co-impacts, not all of which are beneficial. To ensure that policy actions yield co-benefits, the most promising approach appears to be an integration of several policies. In order to select the best and most complementary policies, one must explicitly consider the synergies and trade-offs that may exist between the different technical and institutional options. Policy-makers must also consider the economic, technological and institutional context for the region under consideration – there will be a unique solution for each case. Policies should be assessed both individually and in terms of how they might interact with each other. Evaluation should not be limited to the climate and health impacts of emission reductions and their economic costs, but should be inclusive of noneconomic (social and environmental) costs and co-impacts. It is apparent from this research that certain emission problems could be addressed by alternative fuel use, or choice of engines that are robust to tampering or deterioration. Using CNG fuel in 4stroke engines is a very significant improvement over diesel or 2-stroke engines. Since it is unrealistic to expect all vehicles in a city to convert to CNG, a sound inspection and maintenance program will probably also be necessary to identify high-emitters. When evaluating potential solutions to the challenge of reducing the adverse impacts of motor vehicle emissions, integrated policy-assessment frameworks are likely to reveal important climate/air quality interactions that may not have been otherwise obvious.  124  6.6. References Bennett, D. H.; McKone, T. E.; Evans, J. S.; Nazaroff, W. W.; Margni, M. D.; Jolliet, O.; Smith, K. R. Defining intake fraction. Environ. Sci. Technol. 2002, 36 (9), 206A-211A. Cropper, M. L.; Simon, N. B.; Alberini, A.; Arora, S.; Sharma, P. K. The health benefits of air pollution control in Delhi. Amer. J. Agr. Econ. 1997, 79 (5), 1625-1629. Faiz, A.; Gautam, S.; Gwilliam, K. M. Technical and policy options for reducing emissions from 2-stroke engine vehicles in Asia. Int. J. Veh. Des. 2004, 34 (1), 1-11. Forster, P.; Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R.; Fahey, D.; Haywood, J.; Lean, J.; Lowe, D.; Myhre, G.; Nganga, J.; Prinn, R.; Raga, G.; Schulz, M.; Dorland, R. V. Changes in atmospheric constituents and in radiative forcing, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon, S. et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, NY, 2007. Gouge, B.; Ries, F.; Dowlatabadi, H. Spatial distribution of diesel transit bus emissions and urban populations: Implications of coincidence and scale on exposure. Environ. Sci. Technol. 2010. Kandlikar, M.; Reynolds, C. C. O.; Grieshop, A. P. Alternative perspective on black carbon mitigation as a response to climate change, in Lomberg, B. (ed) Smart Solutions to Climate Change: Comparing Costs and Benefits. Cambridge University Press: Cambridge, UK. (In press). Marshall, J. D.; Riley, W. J.; McKone, T. E.; Nazaroff, W. W. Intake fraction of primary pollutants: Motor vehicle emissions in the South Coast Air Basin. Atmos. Environ. 2003, 37 (24), 3455-3468. Molina, M. J.; Molina, L. T. Megacities and atmospheric pollution. J. Air Waste Manag. Assoc. 2004, 54 (6), 644-680. Pucher, J.; Korattyswaropam, N.; Mittal, N.; Ittyerah, N. Urban transport crisis in India. Transp. Pol. 2005, 12 (3), 185-198. Reynolds, C. C. O.; Kandlikar, M. Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860-5865. Takeuchi, A.; Cropper, M.; Bento, A. The impact of policies to control motor vehicle emissions in Mumbai, India. J. Reg. Sci. 2007, 47 (1), 27-46. Wong, C.; Vichit-Vadakan, N.; Kan, H.; Qian, Z. Public Health and Air Pollution in Asia (PAPA): A Multicity Study of Short-Term Effects of Air Pollution on Mortality. Environ. Health Perspect. 2008, 116 (9), 1195-202. 125  Appendices  126  A. Appendix A: Supporting information for Chapter 2 This Supporting Information document contains: additional details about the public transportation fleet in New Delhi, a discussion of emissions factor development; a summary of global warming/cooling potentials (GWP/GCP) for climate forcing aerosols; and results of a sensitivity analysis for GWP/GCP.  A.1. Supporting information: Background Table A.1. Public transportation vehicles in Delhi. Number of vehicles, average annual activity (km/year), and total annual activity per vehicle category (million km) Public transportation  Number in 2005 a  vehicle type  a b  Average annual activity b  Total annual activity  (km/year)  (million km/year)  Buses  10,480  75,000  786  Cars (Taxis)  16,249  55,000  894  Auto-rickshaws  62,048  44,000  2,730  Total:  88,777  Source: TERI (2006) Source: ISSRC (2004)  Table A.2. Average fuel consumption and CO2 emissions factors. CO2 emissions c  Fuel consumption Public transportation vehicle  Liquid Fuel a  CNG Fuel b  Liquid Fuel  CNG Fuel  type  (kg/100km)  (kg/100km)  (g/km)  (g/km)  Buses (Diesel/CNG)  33.3  41.7  1063  1160  Cars (Gasoline/CNG)  4.9  5.2  157  144  Auto-rickshaws  2.1  2.2  67  62  (Gasoline/CNG) a  Source: Hatwal (2004), ISSRC (2004) CNG buses are assumed to be 25% less fuel efficient than diesel buses; CNG cars and auto-rickshaws are assumed to be 5% less fuel efficient than their gasoline-fueled counterparts. c CO2 emissions factors for each vehicle/fuel type are derived using the assumption that all fuel is completely burned. b  127  A.2. Supporting information: Methods  A.2.1. Emission factors A.2.1.1 Exhaust methane Methane (CH4) is second only to carbon dioxide in importance as a global climate-forcing species (Ramaswamy et al. 2001). It has a GWP of 23, which means that one unit of methane (by mass) released into the atmosphere has the same radiative forcing effect as 23 units of CO2. Natural gas typically contains 85-95% methane by mass, as well as some higher hydrocarbons (ethane, propane, butane), carbon dioxide, nitrogen and trace amounts of other gases. Although the composition of natural gas can vary by season and location of extraction, it is usually controlled during the refining process to avoid problems with engine operation that would otherwise arise in older engines with mechanical fuel-metering technology. In New Delhi, the natural gas contains about 92% methane (Goyal and Sidhartha 2003). Methane is present in the exhaust of retrofitted CNG vehicles due to incomplete combustion of the gaseous fuel-air mixture in engines with suboptimal design of the combustion chamber and the fuel delivery system. Methane is also a component of the mix of hydrocarbon species in the exhaust of liquid-fueled vehicles, although its emissions rate is between one and two orders of magnitude less than from CNG vehicles. We use the following CH4 emissions factors: 0.06, 0.14 and 0.08 g/km for pre-conversion buses, cars and auto-rickshaws respectively; and 6.50, 2.28 and 1.30 g/km for retrofitted CNG buses, cars and auto-rickshaws respectively (Lipman and Delucchi 2002). A.2.1.2 Leaked methane Natural gas leaks from gas compressors at fueling stations, it is spilled during the refueling of CNG vehicles, and it may leak from a vehicle’s onboard fuel-storage and delivery system (referred to as ‘evaporative’ losses in the literature). The emissions associated with natural gas extraction, processing, and transmission to the fueling stations are outside the scope of this study. In 2005, Delhi’s Transport Authority detected CNG leakage from 75% of the buses 128  (EPCA 2005a; EPCA 2005b). This finding led to the implementation of bus-safety policies (in particular related to leakage mitigation) aimed at manufacturers, the transport companies, and the CNG distributors. Mandatory safety checks have been operational since early 2006. Consequently, we assume that CNG leakage from operating vehicles has now been reduced to rates equivalent to industrialized countries. A study conducted on Delhi buses in 2002 documented 10 cases of significant CNG releases involving fires in the previous year (Erlandsson and Weaver 2002), however tank failure in operation is generally considered to contribute negligible amounts of methane compared to exhaust and leakage sources (Delucchi 2003). It is a possibility that lower profile fuel-release events may be more common, but are not reported. Table A.3 illustrates how the total methane ‘leakage’ emissions factors are composed of compressor leakage, refueling leakage, and evaporative leakage (from the fuel storage and distribution system onboard a vehicle). Table A.3. ‘Leakage’ methane emissions factors (g/km) related to compression of the natural gas, vehicle refueling, and ‘evaporative’ emissions from the CNG vehicles’ fuel systems. Methane Emissions Factor (g/km) b Proportion of delivered CNG fuel mass a  Buses  Cars  Autorickshaws  Compressor emissions  0.1 %  0.383  0.048  0.020  Refueling emissions  0.02 %  0.077  0.010  0.004  ‘Evaporative’ leakage  0.4 %  1.533  0.191  0.082  Tank Failure  0%  0  0  0  1.99  0.25  0.11  Total (g/km): a  Source: (Delucchi 2003)  b  Emissions factors take into account the methane content of natural gas, 92%  A.2.1.3 Elemental carbon and organic Carbon Particulate matter (PM) is made up of elemental carbon (EC) and organic carbon (OC) in different ratios depending on its source. PM emitted from heavy-duty diesel vehicles is approximately two-thirds EC (Fraser et al. 2002); from cars it is approximately half EC (Kittelson et al. 2003); and from 2-stroke vehicles it is approximately 5% EC (Sakai et al. 1999). All of the PM emissions factors used in this study, disaggregated into EC and OC components, 129  are given in table 1 (manuscript). Average PM emissions factors depend on the proportion of socalled ‘superemitters’ (Bond et al. 2004). We assume that 40% of buses and cars are superemitters, and that 50% of auto-rickshaws are superemitters. The emissions factors used represent the middle of the range of values presented in the literature for Indian vehicles. The effect of uncertainty in PM emissions factors is investigated in section 2.3.2 in the manuscript (Chapter 2). Laboratory engine testing show that total PM mass emissions from purpose-built CNG engines are about 5% that of a new diesel engine (McTaggart-Cowan et al. 2006). Few studies have examined in-use PM emissions from retrofitted CNG vehicles. We use data presented in Sanghi (2001) for buses, and ISSRC (2004) for cars and auto-rickshaws. Uncertainty in PM emissions factors for CNG vehicles has an almost negligible impact on the total change in CO2-eq emissions because they are low relative to the PM emitted from pre-conversion vehicles.  A.2.1.4 Sulphate Sulphate mass emissions are proportional to the fuel sulfur content. In New Delhi, the Bharat Stage III Norms (Indian vehicle emissions and fuel standards approximately equivalent to European ‘Euro 3’) have applied since 2005. These regulations mandate that the sulfur content must be less than 350ppm for diesel, and less than 150ppm for gasoline (CPCB 2007; DieselNet 2007). The sulfur content in natural gas is negligible. The emissions factors for sulfur dioxide (SO2), which is the gaseous precursor to sulphate particulate, are readily calculated from the fuel sulfur content and the fuel consumption of the different vehicle types.  130  A.2.2. Global warming/cooling metrics for climate forcing aerosols Table A.4. Global warming/cooling potentials Vehicle emission  Global Warming/Cooling Potential (GWP/GCP)  Source  a  Carbon Dioxide  CO2  1  Ramaswamy et al. (2001)  Methane  CH4  23  Ramaswamy et al. (2001)  Elemental Carbon  EC  455 (193 to 716)  (This study)  Organic Carbon  OC  -35 (-9 to -83)  (This study)  Sulfur Dioxide  SO2  -100 (-46 to -213)  (This study)  a  100 year time horizon; uncertainty ranges in parentheses, calculated from uncertainty in radiative forcing values (Forster et al. 2007)  Table A.5. Total annual climate-forcing emissions inventory. Units are 1000 tons of CO2-equivalent (CO2-eq) emissions, attributable to the public transportation fleet before (liquid-fuel vehicles) and after (CNG-fuel vehicles) the fuel-switching event. Emissions species  CO2-eq Emissions (1000 tons)  CO2-eq Emissions (1000 tons)  Liquid-Fuel Vehicles  CNG Vehicles Auto-  Auto-  Buses  Cars  rickshaws  Buses  Cars  rickshaws  Carbon Dioxide  835.4  140.0  183.3  912.0  129.0  168.9  Methane (exhaust)  1.1  2.8  4.9  117.5  46.9  81.6  Methane (leakage)  0  0  0  36.0  5.1  6.7  Subtotal GHGs:  836.5  142.8  188.2  1065.5  181.0  257.2  Elemental Carbon  543.2  65.0  12.4  0.6  0.4  9.8  Organic Carbon  -13.3  -5.4  -18.3  -0.1  -0.1  -2.3  SO2 (precursor to sulphate)  -18.3  -1.3  -1.7  0  0  0  Subtotal Aerosols:  511.6  58.3  -7.6  0.5  0.3  7.5  Total CO2-eq emissions:  1348  201  181  1066  181  265  Gaseous (GHGs):  Aerosols:  131  A.3. Supporting information: Results  Figure A.1. Sensitivity of model results. GWP of EC is varied from its lower (GWP = 193) and upper (GWP = 716) 95% confidence bounds. The x and y axes refer to PM and CH4 emissions factors respectively. The contours on the graph indicate the value of the change in CO2-eq before and after the CNG switch, for the range of emissions factors shown (‘low’, ‘medium’ and ‘high’ values), and the shading shows the area of net climate benefit (CO2-eq reduction).  132  A.4. Supporting information: References Bond, T. C., et al. A technology-based global inventory of black and organic carbon emissions from combustion. J. Geophys. Res. 2004, 109, (D14203). CPCB. Indian Diesel & Gasoline Specifications. Central Pollution Control Board: Delhi, India, 2007: http://cpcb.nic.in/Environmental%20Standards/Auto_Fuel_Quality.html. Delucchi, M. A., Lifecycle Emissions Model (LEM): Lifecycle emissions from transportation fuels, motor vehicles, transportation modes, electricity use, heating and cooking fuels, and materials. (Documentation of methods and data). Institute of Transportation Studies, University of California, Davis: Davis, CA, 2003: http://www.its.ucdavis.edu/people/faculty/delucchi/index.php. DieselNet. Emission Standards India: On-Road Vehicles and Engines. 2007: http://www.dieselnet.com/standards/in/. EPCA. Assessment and prevention of gas leakage from CNG buses. Environment Pollution (Prevention & Control) Authority for the National Capital Region: New Delhi, India, 2005: http://www.cpcb.nic.in/. EPCA. Supplementary submission on the safety issues in CNG buses. Environment Pollution (Prevention & Control) Authority for the National Capital Region: New Delhi, India, 2005: http://www.cpcb.nic.in/. Erlandsson, L.; Weaver, C. Safety of CNG buses in Delhi: Findings and recommendations. Centre for Science and Environment: New Delhi, India, 2002: http://www.cseindia.org. Forster, P., et al. Changes in atmospheric constituents and in radiative forcing, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, S. Solomon, et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, 2007. Fraser, M. P., et al. Variation in composition of fine particulate emissions from heavy-duty diesel vehicles. J. Geophys. Res.-Atmos. 2002, 107, (D21). Goyal, P.; Sidhartha. Present scenario of air quality in Delhi: a case study of CNG implementation. Atmos. Environ. 2003, 37 (38), 5423-5431. Hatwal, S. C. The contribution of natural gas towards cleaner air in India, in BAQ 2004. CAIASIA: Agra, India, 2004: http://www.cleanairnet.org/baq2004/. ISSRC. Development of the base emissions rates for the International Vehicle Emissions Model, Appendix A: Base emissions rates. International Sustainable Systems Research Centre: La Habra, CA, 2004: http://www.issrc.org/.  133  ISSRC. Pune vehicle activity study (March 9-March 22 2003). International Sustainable Systems Research Centre: La Habra, CA, 2004: http://issrc.org/. Kittelson, D. B., et al. Gasoline vehicle exhaust particle sampling study, in 9th Diesel Engine Emissions Reduction (DEER) Workshop 2003. Newport, RI, 2003: http://www.osti.gov/bridge/product.biblio.jsp?osti_id=829821. Lipman, T. E.; Delucchi, M. A. Emissions of nitrous oxide and methane from conventional and alternative fuel motor vehicles. Climatic Change. 2002, 53 (4), 477-516. McTaggart-Cowan, G.P.; Reynolds, C.C.O.; Bushe, W.K. Natural gas fuelling for heavy-duty on-road use: current trends and future direction. Int. J. Environ. Stud. 2006, 63 (4), 421-440. Ramaswamy, V., et al. Radiative Forcing of Climate Change, in Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change, J. T. Houghton, et al., Editors. Cambridge University Press: Cambridge, United Kingdom and New York, 2001, 349-416. Sakai, T.; Nakajima, T.; Yamazaki, H. O-PM/Emitted matters caused by two-stroke engine oil and its reduction. SAE Tech. Pap. 1999, 1999-01-3260. Sanghi, S.; Kale, S.R.; Mohan, D. Air quality impact assessment by changeover to CNG buses in Delhi. Transportation Research and Injury Prevention Programme, IIT Delhi: Delhi, 2001: http://web.iitd.ac.in/~tripp/rp/report/cng_changover_ioc.pdf. TERI, TERI Energy Data Directory and Yearbook 2004/05. The Energy and Resources Institute: New Delhi, India, 2006.  134  B. Appendix B: Supporting information for Chapter 3  B.1. UBC Behavioural Research Ethics Board: Approval/Renewal certificates  135  136  B.2. Survey consent form for auto-rickshaw drivers (English and Hindi)  fo"k; vkbZMh uEcj % Survey of Vehicle Operators in India Consent Form Investigators: � Milind Kandlikar, Institute for Asian Research and Liu Institute for Global Issues, University of British Columbia, Canada, phone: ++1-604-8226722. � Madhav Badami, School of Urban Planning and School of Environment, McGill University, Canada; phone: ++1-514-3983183. � Geetam Tiwari, Transportation Research and Injury Prevention Programme (TRIPP), Indian Institute of Technology Delhi, India; phone: ++91-11-26596361, 26858703. � Conor Reynolds, Institute for Resource, Environment and Sustainability, University of British Columbia, Canada. This research is part of the graduate thesis project of Conor Reynolds, a Doctoral student at the University of British Columbia. Conor is working under the supervision of Dr. Milind Kandlikar.  What is a Consent Form?  We are asking you to be in a research study. The purpose of this consent form is to give you information to help you decide whether or not to participate. You may ask questions about anything that is not clear. When all of your questions have been answered, you can decide if you want to be in this study or not. This process is called "informed consent." We will give you a copy of this form for your records.  What is the purpose of this study?  We are studying public transport vehicles in New Delhi and other Indian cities. We want to learn more about the use patterns, fuel consumption, maintenance requirements and running costs for these vehicles, by talking with the people who know the vehicles best – the vehicle operators. The study is funded by the Canadian Social Studies and Humanities Research Council (SSHRC), but is being conducted independently by the researchers named above. The results of this survey will be used to better understand the needs of public transport vehicle operators, and may be used by decision makers in the government who plan changes to the Indian public transportation systems.  What are the study procedures?  You have been invited to participate because you are public transport vehicle operator. To conduct this research, we need to ask you questions about the type of vehicle you drive; about operating expenses and the costs of owning or renting; about the trips you took on the previous day; and details about fueling and maintenance. The interview will take about half an hour. We will be conducting the same interview for about 400 people.  Is participation voluntary?  Yes. You may choose not to be in this study and you may choose not to participate in any part of the study. You may decline to participate or withdraw from the study at any time, without any penalty or loss of benefits to which you are otherwise entitled.  How will the information be protected?  All the questionnaires will be coded with a number, not your name. We will keep a record of the number we assign to you, but we will keep that separate from any other information we collect, and we will not release the code to anyone else.  Who will have access to the data?  Only the researchers and their research assistants will have access to information about you. Study information identifying you will not be revealed to anybody else in Canada or India. Your name will not be used in any published reports about this study.  Remuneration/Compensation:  In order to defray the costs of loss of wages, each person who participates in this study will receive an honorarium in the amount of Rs.100.  Questions?  If you have any questions about this study, please call one of the people listed on the front page of this consent form.  10-12-2008 ----------------------  --------------------------------------------------------------------  Signature of Researcher, Dr. Milind Kandlikar  Date  Consent to participate: Your name below indicates that you consent to participate in this study, and that you have received a copy of this consent form for your own records.  ------------------------------------------  Subject Name  ----------------------------------------  Auto License-plate number  -------------------Date  If you are interested in being part of future studies, please provide your phone number so we can contact you: Phone #: _____________________________  137  fo"k; vkbZMh uEcj % vè;;u lewg %  vkWVkspkydksa dk losZ lgefr&i=k  · fefyUn dkUMyhdj] baLVhV~;wV iQkWj ,f'k;u fjlpZ ,.M Y;w baLVhV~;wV iQkWj Xykscy b';wfll] ;wfuoflZVh vkWiQ fczfV'k dksyfEc;k] dukMk] nwjHk"k % $$1&604&8226722 · ek/ko cknkeh] Ldwy vkWiQ vjcu Iykfuax ,.M Ldwy vkWiQ ,Wuok;juesUV] ,elhfty ;wfuoflZVh] dukMk] nwjHkk"k % $$1&514&3983183 · xhre frokjh] VªakUliksVZs'ku fjlpZ ,.M bUT;qjh fizosUlu izksxzke (fVªi)] bafM;u bafLV~V;wV vkWiQ VsDuksyksth] fnYyh] Hkkjr nwjHkk"k % $$91&11&26596361] 26858703 · dkWukj jsukWYM~l] bafLV~V;wV iQkWj fjlkslZ] ,Wuok;juesUV ,.M lLVsusfcfyVh] ;wfuoflZVh vkWiQ fczfV'k dksyfEc;k] dukMk A ;g 'kks/ dkukWj jsuksYM~l ds }kjk fd, tk jgs xzstq,V Fksfll dk ,d Hkkx gS] tks ;wfuoflZVh vkWiQ fczfV'k dksyafc;k 'kks/ ds Nk=k gS A dkukWj jsuksYM~l ;g 'kks/ Mk- fefyUn dkWUMyhdj dh fuxjkuh esa dj jgs gSa A  lgefr&i=k D;k gS \  ge vkidks bl 'kks/&vè;;u esa bUVjO;w nsus ds fy, vkeaf=kr dj jgs gSa A bl lgefr&i=k dk mn~ns'; 'kks/ ds ckjs esa lwpuk nsuk gS ftlls vki r; dj lds fd vkidks bUVjO;w nsuk gS ;k ugha A vki bl 'kks/ ls lEcfU/r dksbZ Hkh iz'u iwN ldrs gSa ftlls vkidh 'kadk nwj gks tk; A tc vkids lHkh iz'uksa ds mRrj fey tk,a rks vki fu.kZ; ys ldrs gS fd vki bl 'kks/&vè;;u esa Hkkx ysaxs ;k ugha A bls ^lgefr&i=k* dgrs gSa A ge] vkidks vius fjdkWMZ ds fy, bl iQkeZ dh ,d izfrfyi nsaxs A  bl vè;;u dk mn~ns'; D;k gS \  ge Hkkjr ds vU; 'kgjksa vkSj ubZ fnYyh esa lkoZtfud ifjogu ds lk/uksa dk vè;;u dj jgs gSaA ge] vkWVkspkydksa (tks vkVksfjD'kk vkSj mlds pkyd dks vPNh rjg tkurs gS) ls ckr djds iz;qDr rjhdksa] bZa/u&[kir] t:jh j[k&j[kko vkSj vkWVksfjD'kkvksa ij vkus okyh ykxrksa ds ckjs esa tkudkjh pkgrs gSa A bl vè;;u ds fy, dukfM;u lks'ky LV~Mhl ,.M â;wesfuVht fjlpZ dkWfmUly (,l-,l-,p-vkj-lh-)] }kjk vkfFkZd enn nh xbZ gS ij ;g 'kks/&vè;;u mijksDr fyf[kr 'kks/drkZvksa }kjk Lo;a vk;ksftr fd;k x;k gSA bl losZ dk ifj.kke vkWVkspkydksa dh t:jrksa dks vPNh rjg le>us ds fy, mi;ksx esa yk;k tk,xk A bldk iz;ksx ljdkj ds fdlh foHkkx }kjk ugha fd;k tk,xk ftlls vki yksx izHkkfor gksa A  vè;;u ds rjhds D;k gSa \  vkWVksfjD'kk lkoZtfud ifjogu dk ,d fgLlk gS A bl 'kks/ ds fy, gesa vkils vkWVksfjD'kk ds ckjs esa iz'u iwNus dh vko';drk gS & vkWVksfjD'kk ds izdkj ds ckjs esa] vkWVksfjD'kk pykus esa vk, [kpsZ] [kqn [kjhnus ;k fdjk, ij ysus dh ykxr] fVªi ds ckjs esa tks vkius fiNys fnu fy, vkSj bZa/u rFkk j[k&j[kko dk iw.kZ fooj.k A lk{kkRdkj esa yxHkx vk/k ?kaVk yxsxk A ge ;g lk{kkRdkj yxHkx 350 yksaxksa ls djsaxs A  D;k Hkkxhnkjh LoSfPNd gS \  vki bl vè;;u esa ^gkW * ;k ^u* pqu ldrs gS vkSj vki fgLlk ysus ds fy, euk Hkh dj ldrs gS A  lwpuk lqjf{kr dSls j[kh tk,xh \  lHkh iz'uksa ds tokc dks ^dksM* ds ekè;e ls n'kkZ;k tk,xk A bl 'kks/ esa fdlh O;fDr ds uke dk ftØ ugha gksxkA gj O;fDr dks ^vadksa* ds vk/kj ij tkuk tk;sxkA tks vad vkidks fn, tk,xsa mudk ge fjdkWMZ j[ksaxs bl fjdkMZ dh tkudkjh fdlh vU; O;fDr dks ugha nh tk;sxh A  MkVk dks dkSu ns[k ldsxk \  vkidh lwpuk dsoy 'kks/drkZ vkSj muds lgk;d gh ns[k ldsaxsA vkids }kjk ,df=kr dh xbZ 'kks/ lwpuk fdlh dks ugh nh tk,xh pkgs og Hkkjr esa jgrk gks ;k dukMk esa A bl vè;;u ds fdlh Hkh jiV esa vkids uke dk ftØ ugha fd;k tk,xk A  Hkqxrku@vnk;xh %  vkids etnwjh dk uqdlku ugh gks blds fy, vkidks 100 #i;s HkRRkk esa fn;k tk,xk A  ç'u %  ;fn vkidks bl vè;;u ds ckjs esa dksbZ ç'u iwNuk gS rks ÑIk;k bl lgefr i=k ds izFke ist ij nh xbZ lwph esa ls fdlh Hkh O;fDr ls iQksu djds iwN ldrs gSa A 10-12-2008 ---------------------fnukad  -------------------------------------------------------------------'kks/drkZ ds gLrk{kj] Mk- fefyUn dkWUMyhdj  Hkkx ysus dh lgefr%  vkids gLrk{kj n'kkZrs gS fd bl v?;;u esa Hkkx ysus dh vkidh lgefr gS] vkSj vkius bl lgefr&i=k dh ,d izfr vius fjdkWMZ ds fy, j[k yh gS A  -----------------------------------------O;fDr dk uke  ------------------------------------vkWVks ykblsal IysV uEcj  -------------------fnukad  ;fn vki Hkfo"; esa gksus okys vè;;u esa 'kkfey gksus ds bPNqd gS rks Ñi;k viuk iQksu uEcj fy[ksa rkfd ge vkidks lEidZ dj ldsA iQksu uEcj % --------------------------------------------  138  B.3. Survey instrument for ‘Three-Seater Rickshaw’ (TSR) drivers  (Survey instrument begins next page)  139  140  141  142  143  144  145  146  147  148  149  150  C. Appendix C: Supporting information for Chapter 4  C.1. Emissions regulations and test vehicles In response to public concerns about air quality and health, public transit vehicles in Delhi were required to switch from gasoline or diesel to compressed natural gas (CNG) fuel. The Indian Supreme court made this unique ruling outside the purview of Indian regulatory bodies. However, India does have a regulatory framework to control new and in-use vehicle emissions. Since the early 1990s, all new vehicles sold in India have been required to meet mass emission limits (called the Bharat Stage Norms) for certain exhaust constituents measured over the Indian Drive cycle (DieselNet 2010). Bharat Stage Norms regulate emissions and fuels for fourwheeled vehicles (cars, buses, trucks), similar to European regulations, but also regulate twoand three-wheeled vehicles. The Bharat Stage III Norms (currently in force for new threewheelers) impose emission limits of 1.25 g km-1 for CO and 1.25 g km-1 for the sum of THC+NOX (Table C.1). Recognizing that emission control devices deteriorate and pollutant emissions tend to increase with vehicle age, all in-use Indian vehicles must pass bi-annual emission inspections called Pollution Under Control (PUC), during which the concentration of CO and THC are measured in the undiluted exhaust stream from the idling engine (DOT 2010). Auto-rickshaw drivers must hold a valid PUC certificate in order to operate legally. However PUC (and its equivalent in other parts of Asia) has been broadly criticized for its ineffectiveness, and is urgently in need of reform (Rogers 2002; Faiz et al. 2006). Problems with PUC include: NOX, CH4 and PM are not measured (Rogers 2002), ‘test & repair’ garages are prone to corruption (Rogers 2002; Hausker 2004), and idle CO and THC measurements are generally not well correlated with emissions during driving (Mazzoleni et al. 2004). An improved system has been proposed that would involve chassis dynamometer testing over a simple drive-cycle at central ‘test-only’ locations (Hausker 2004).  151  Table C.1. Indian mass emission limits for three-wheeled vehicles, as measured on the Indian Drive Cycle. As of April 2010, Bharat Stage IV Norms have been introduced in 11 major cities across India. Bharat Stage IV is equivalent to Euro 4 (implemented in Europe in 2005). Year Standard CO (g km-1) THC+NOX (g km-1) PM (g km-1) Three-wheelers with spark-ignition engines fueled by gasoline, CNG, or LPG (liquefied petroleum gas) 1996 6.75 5.40 2000 4.0 2.0 2005.04 Bharat Stage II 2.25 2.0 2005.04 Bharat Stage III 1.25 1.25 Three-wheelers with diesel engines 2005.04 Bharat Stage II 1.0 0.10 2005.04 Bharat Stage III 0.5 0.05  Table C.2. Specifications of spark-ignited auto-rickshaws fueled with compressed natural gas (CNG) or gasoline/petrol (PET), with rrear-mounted engines in either 2-stroke or 4-stroke configurations. Fuel / engine type  CNG 4-stroke  Gasoline (‘petrol’) 4-stroke  CNG 2-stroke  Notation for group  CNG-4S  PET-4S  CNG-2S a  No. of vehicles tested  17 (9 ‘new’, 8 ‘old’)  11 (7 ‘new’, 4 ‘old’)  Manufacturer  Bajaj Auto Limited  Bajaj Auto Limited  Bajaj Auto Limited  Fuel tank capacity  4 kg (~30 liter tank)  3 liter ‘back-up’ tank  4 kg (~30 liter tank)  Displacement  173.5 cc (one-cylinder)  173.5 cc (one-cylinder)  Max power  4.8 kW @ 5000 rpm  6.0 kW @ 5000 rpm  13  b  145.5 cc (one-cylinder) 4.9 kW @ 5000 rpm  b  Max torque  9.3 Nm @ 2500 rpm  11.5 Nm @ 4000 rpm  10.8 Nm @ 4000 rpm  Gross vehicle weight  360 kg c  360 kg c  335 kg  Max payload  325 kg  325 kg  325 kg  Design no. of passengers (not including driver)  3  3  3  a  All of these vehicles were also tested both gasoline and CNG mode, allowing a comparison of emission factors between the groups. b Max power and torque values given here are for dedicated gasoline-fueled auto-rickshaws; since the vehicles tested in this study were tuned to operate on CNG as their primary fuel, their performance may differ. c Dedicated gasoline vehicles have a slightly lower gross vehicle weight (317kg) than the dual-fuel vehicles tested in this study, because gasoline-only vehicles do not have the additional mass of a CNG fueling system. Source: Bajaj (2010)  152  C.2. Supporting information: Methods  C.2.1. Details of test protocol and emission measurements Vehicles were tested on a 70kW single-roller AVL Zollener chassis dynamometer. The vehicle exhaust was connected via a stainless steel line to a constant volume sampler (CVS, model: AVL CVS-4000) to dilute the vehicle exhaust. Total flow rate in the CVS was varied from vehicle-tovehicle (between 1-10 cubic meters per minute [m3 min-1], in 1 m3/min increments) to maintain diluted PM concentrations of less than approximately 100 mg/m3 during warm up cycles. Test average dilution ratios varied between 15 and 40 during tests, though the instantaneous dilution ratio has a much greater range. A professional driver from ICAT drove the test vehicles. Each vehicle was warmed up for approximately 10 minutes both on-road and then on the chassis dynamometer before data collection, therefore cold-start emissions were not obtained during this study. The decision to omit cold-start emissions was made because auto-rickshaws (like all taxis) operate for most of the day and hence cold-starts are rare; in addition there was the practical reason that test vehicles were only available for the day of testing, making an overnight cold-soak impossible. We used the 648-second duration Indian Drive Cycle (IDC), which has an average speed (including idle time) of 21.9 km h-1, top speed of 42 km h-1, and total travel distance of 3.9 km (Figure C.1 and Table C.3). The proportion of time spent idling, at steady speed, accelerating and decelerating was 15%, 12%, 39% and 34%, respectively. For gaseous pollutant measurements, diluted engine exhaust was drawn into Tedlar bags from the CVS at 10 l min-1 over the duration of the test; a simultaneous sample of test-cell air was collected for background corrections. Gaseous pollutants in the integrated samples (CO, NOX, THC, CH4 and CO2) were measured immediately after completion of the drive cycle to determine the test-average emissions using an AVL-Pierburg AMA–4000 analyzer bench. CO and CO2 emissions were measured using non-dispersive infrared analyzers (limit of detection [LOD] = 50 ppb), THC and CH4 were measured with a flame ionization detector (LOD = 6 ppb), and NOX was measured with a chemi-luminescence analyzer (LOD = 6ppm). Uncertainty for gas  153  analyzers was assumed to be ±5%. A PUC idle emission test of undiluted tailpipe CO and THC (AVL Digas 444) was performed after completion of the IDC. PM2.5 samples were collected on two filter trains, each consisting of a PM2.5 cyclone followed by a 47mm diameter filter-holder. One filter train – denoted ‘Bare-Q’ – held a pre-fired quartz filter (Pall Tissuquartz), the other held a pre-weighed Teflon filter (Pall Teflon, 2 micron pore size) followed by a backup quartz filter (‘quartz behind Teflon’, QBT) used to correct for positive sampling artifact in Bare-Q (Subramanian et al. 2004). The sample flow rate through each filter train was 20 l min-1. In addition to the test samples, four dynamic blank samples (filters that sampled only dilution air for the test duration) and four handling blanks were collected; approximately one of each for every 10 vehicle tests. After testing, the Teflon filters were reweighed under controlled conditions (T = 23±3°C, RH = 40±5%) to determine PM mass. Organic carbon/elemental carbon (OC/EC) analysis was conducted on the quartz filters using a modified version of the NIOSH 5040 thermal-optical transmittance (TOT) protocol in a Sunset Laboratory OC/EC Analyzer (NIOSH 2003). OC measured on the Bare-Q filter was corrected for positive sampling artifact by subtracting OC on the QBT filter (Subramanian et al. 2004). The gas-particle partitioning of OC emitted by engines has been shown to be a function of the concentration of organic aerosol (COA) during sampling (Robinson et al. in press), with greater partitioning into the particle phase at higher COA. For our measurements, this effect may bias OC emission factors by up to a factor of ~4 relative to atmospheric conditions. However, our samples were collected under conditions similar to those in other emission tests (ARAI 2007) that were relatively constant across vehicle classes (COA values are given for individual tests in Table C.5), enabling valid comparison among our vehicle groups and between our data and other studies. The impact of phase partitioning on our emission data will be examined in more detail in a companion paper on OC phase partitioning and speciation, in preparation.  C.2.2. Details regarding data analysis Distance-based mass emission factors (EFi, grams of species i per kilometer, g km-1) were calculated for gaseous species (CO, THC, NOX, CH4, CO2) using Equation C.1.:  154  EFi =  !  VCVS " # i " k H " (Csample,i $ Cambient,i )" 10 $3 d  (Equation C.1.)  where VCVS is the total CVS flow (m3 per test, corrected to standard conditions of 293 K and 101.3 kPa), !i is density of the species i (kg m-3) at standard conditions, kH = humidity correction factor (for NOX only) (Gingrich et al. 2003), Csample and Cambient are species mixing ratios (ppm) in the diluted exhaust gas and dilution air, respectively, and d is the drive cycle distance (km). Vehicle fuel consumption was not measured directly during testing, but was calculated using a carbon balance of the major carbon-containing exhaust constituents: CO2, CO, and THC (the latter included CH4) (Kirchstetter et al. 1999). Automotive fuel CNG specifications have been proposed (EPCA 2007) but actual composition of CNG in Delhi was unavailable. Indian gasoline is unleaded with an octane rating of 91 in urban areas, and sulfur content is limited to 0.015% by weight (APEC 2010; CPCB 2010). The carbon weight fractions for CNG and gasoline were assumed to be 0.76 and 0.87, respectively. PM emissions contributed negligibly to overall carbon emissions in the fuel-consumption calculation; PM carbon contributed less than 0.35% of total emitted carbon during 4-stroke tests, while lubricating oil (rather than fuel) is the primary source of PM emissions from 2-stroke vehicles (Volckens et al. 2007; Kojima et al. 2002). Fuel consumption is reported in units of mass of fuel consumed per distance traveled (kg 100km-1); mass based fuel consumption estimates for CNG and gasoline are roughly equivalent because the energy density of the two fuels are estimated to be within 5% of each other (approximately 46 MJ kg-1).  155  Figure C.1. Sub-cycle from the Indian Drive Cycle (IDC); this sub-cycle lasts 108 seconds, including idling, and is repeated 6 times to make up the complete IDC.  156  Table C.3. Velocity table for one 108-second sub-cycle of the Indian Drive Cycle. Time (s) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27  Target Velocity (m s-1) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.65 1.3 1.95 2.6 3.25 3.9 4.46 5.02 5.58 6.14 5.51  Time (s) 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54  Target Velocity (m s-1) 4.88 4.25 3.62 3.62 3.62 4.18 4.74 5.3 5.86 6.42 6.86 7.3 7.74 8.18 8.62 8.06 7.5 6.94 6.94 6.94 6.94 6.94 6.38 5.82 6.27 6.72 7.17  Time (s) 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81  Target Velocity (m s-1) 7.62 8.07 8.52 8.97 9.42 9.74 10.06 10.38 10.7 11.02 11.34 11.66 11.2 10.74 10.28 10.28 10.28 10.28 10.28 10.28 10.28 10.28 9.86 9.44 9.76 10.08 10.4  Time (s) 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108  Target Velocity (m s-1) 10.72 11.04 11.36 11.68 11.22 10.76 10.3 9.84 9.38 8.92 8.46 8 7.54 7.02 6.5 5.98 5.46 4.94 4.42 3.9 3.34 2.78 2.22 1.66 1.1 0.54 0  157  C.3. Supporting information: Results Table C.4. Fuel-based emission factors for the 41 vehicle tests in this study. Expressed uncertainties in EFs are the result of error analysis based on instrument uncertainties. Test No., Vehicle No., Model Year CNG-4S (New) T01-V01-2008 T04-V02-2009 T05-V03-2007 T07-V04-2008 T19-V11-2009 T23-V14-2009b T25-V15-2008 T36-V26-2007 T38-V27-2008 CNG-4S (Old) T09-V05-2001 T11-V06-2001 T13-V07-2001 T14-V08-2001 T16-V09-2000 T17-V10-2000 T20-V12-2001 T21-V13-2001 PET-4S (New) T02-V01-2008 T03-V02-2009 T06-V03-2007 T18-V11-2009 T22-V14-2009b T24-V15-2008 T37-V27-2008  Fuel Cons. (kg 100km-1)  CO (g kg-1)  HC (g kg-1)  NO (g kg-1)  CH4 (g kg-1)  CO2 (g kg-1)  PM (g kg-1)  OC (g kg-1)  EC (g kg-1)  COA a (mg m3)  1.88 ± 0.11 2.08 ± 0.12 2.45 ± 0.14 2.16 ± 0.12 1.99 ± 0.11 1.79 ± 0.1 2.18 ± 0.12 1.81 ± 0.11 2.14 ± 0.12  18 ± 1 168 ± 13 68 ± 5 127 ± 10 42 ± 3 41 ± 3 58 ± 4 42 ± 3 108 ± 8  26 ± 2 53 ± 4 86 ± 7 57 ± 4 119 ± 9 101 ± 8 100 ± 8 12 ± 1 117 ± 9  16.5 ± 1.3 19.7 ± 1.5 17.9 ± 1.4 25.5 ± 1.9 27.1 ± 2 23.5 ± 1.8 29.3 ± 2.2 34.3 ± 2.7 40.2 ± 3  26 ± 2 59 ± 4 78 ± 6 48 ± 4 77 ± 6 97 ± 7 72 ± 5 19 ± 2 84 ± 6  2810 ± 220 2480 ± 190 2550 ± 190 2540 ± 190 2580 ± 190 2580 ± 200 2570 ± 190 2830 ± 220 2420 ± 180  (n/a) 0.035 ± 0.004 0.66 ± 0.04 0.95 ± 0.05 0.039 ± 0.004 0.023 ± 0.002 2.58 ± 0.14 0.063 ± 0.006 0.5 ± 0.03  (n/a) 0 ± 0.01 0.32 ± 0.03 0.57 ± 0.05 0.025 ± 0.008 0.018 ± 0.005 1.49 ± 0.11 0.034 ± 0.011 0.27 ± 0.02  (n/a) 0.016 ± 0.003 0.134 ± 0.011 0.148 ± 0.013 0.009 ± 0.003 0.006 ± 0.002 0.39 ± 0.03 0.008 ± 0.003 0.141 ± 0.012  (n/a) 0 ± 0.03 1.4 ± 0.1 2.25 ± 0.14 0.09 ± 0.03 0.12 ± 0.03 6 ± 0.3 0.12 ± 0.04 1.07 ± 0.08  2.03 ± 0.12 2.19 ± 0.12 2.7 ± 0.15 2.31 ± 0.14 2.54 ± 0.15 2.1 ± 0.12 1.99 ± 0.12 1.92 ± 0.11  28 ± 2 195 ± 15 223 ± 17 14 ± 1 89 ± 7 148 ± 11 17 ± 1 31 ± 2  100 ± 8 22 ± 2 66 ± 5 13 ± 1 39 ± 3 11 ± 1 16 ± 1 39 ± 3  31 ± 2.4 12.3 ± 0.9 7.8 ± 0.6 14.7 ± 1.2 19.3 ± 1.5 12.4 ± 1 13.6 ± 1.1 32.8 ± 2.6  75 ± 6 32 ± 2 55 ± 4 9±1 31 ± 2 12 ± 1 20 ± 2 59 ± 5  2570 ± 200 2530 ± 190 2380 ± 180 2860 ± 230 2660 ± 200 2660 ± 200 2860 ± 230 2770 ± 220  0.34 ± 0.02 0.91 ± 0.05 0.35 ± 0.02 0.58 ± 0.03 0.43 ± 0.02 0.039 ± 0.004 0.09 ± 0.01 1.02 ± 0.06  0.148 ± 0.015 0.73 ± 0.06 0.165 ± 0.016 0.34 ± 0.03 0.17 ± 0.016 0.019 ± 0.008 0.044 ± 0.009 0.51 ± 0.04  0.126 ± 0.011 0.081 ± 0.01 0.062 ± 0.006 0.024 ± 0.004 0.11 ± 0.01 0.011 ± 0.003 0.009 ± 0.003 0.21 ± 0.02  0.55 ± 0.05 1.5 ± 0.1 0.75 ± 0.06 1.4 ± 0.1 0.8 ± 0.06 0.07 ± 0.03 0.16 ± 0.03 1.8 ± 0.1  3.36 ± 0.16 3.92 ± 0.2 4.25 ± 0.2 4.11 ± 0.19 2.34 ± 0.14 3.23 ± 0.16 3.62 ± 0.17  678 ± 47 565 ± 40 1003 ± 69 649 ± 44 76 ± 6 814 ± 57 972 ± 67  181 ± 13 71 ± 5 219 ± 15 313 ± 21 60 ± 5 63 ± 4 182 ± 13  12.2 ± 0.8 15.8 ± 1.1 1.6 ± 0.1 12.4 ± 0.8 27.3 ± 2.1 15.8 ± 1.1 3.9 ± 0.3  12.5 ± 0.9 11.5 ± 0.8 12.7 ± 0.9 8 ± 0.5 0.4 ± 0 9.9 ± 0.7 14.6 ± 1  1740 ± 120 2220 ± 160 1110 ± 80 1440 ± 100 3050 ± 230 1900 ± 130 1300 ± 90  0.69 ± 0.03 0.52 ± 0.03 1.46 ± 0.06 0.8 ± 0.03 0.017 ± 0.003 3 ± 0.14 0.66 ± 0.03  0.31 ± 0.02 (n/a) 0.65 ± 0.05 0.162 ± 0.013 0 ± 0.01 1.77 ± 0.12 0.28 ± 0.02  0.26 ± 0.02 (n/a) 0.48 ± 0.03 0.5 ± 0.03 0 ± 0.002 0.44 ± 0.03 0.28 ± 0.02  3.8 ± 0.2 (n/a) 5 ± 0.3 1.2 ± 0.1 0.02 ± 0.03 7.1 ± 0.4 1.3 ± 0.1  (Table C.4. continues on next page)  158  Table C.4. (continued) Test No., Vehicle No., Model Year PET-4S (Old) T08-V05-2001 T10-V06-2001 T12-V07-2001 T15-V08-2001 CNG-2S (all) T26-V16-1998 T27-V17-2000 T28-V18-2000 T29-V19-1999 T30-V20-2000 T31-V21-1998 T32-V22-1999 T33-V23-1998 T34-V24-2001c T35-V25-1998 T39-V28-2000 T40-V29-1998 T41-V30-1998 T42-V31-1999 a  Fuel Cons. (kg 100km-1)  CO (g kg-1)  HC (g kg-1)  NO (g kg-1)  CH4 (g kg-1)  CO2 (g kg-1)  PM (g kg-1)  OC (g kg-1)  EC (g kg-1)  COA a (mg m3)  4.83 ± 0.23 3.29 ± 0.16 4.31 ± 0.22 4.59 ± 0.21  1191 ± 83 1082 ± 77 1325 ± 94 693 ± 47  243 ± 17 21 ± 1 47 ± 3 334 ± 23  0.2 ± 0 3.3 ± 0.2 0.5 ± 0 9.6 ± 0.7  15.7 ± 1.1 11 ± 0.8 9.1 ± 0.6 7.4 ± 0.5  750 ± 50 1530 ± 110 1090 ± 80 1300 ± 90  2.15 ± 0.34 1.48 ± 0.07 0.63 ± 0.03 1.95 ± 0.08  1.05 ± 0.07 0.92 ± 0.07 0.19 ± 0.02 0.7 ± 0.05  0.53 ± 0.04 0.23 ± 0.02 0.24 ± 0.02 0.82 ± 0.06  6.2 ± 0.3 2.8 ± 0.2 0.8 ± 0.1 3 ± 0.2  2.53 ± 0.13 2.16 ± 0.11 2.05 ± 0.11 2.72 ± 0.14 2.22 ± 0.11 2.75 ± 0.14 1.96 ± 0.11 3.06 ± 0.15 (n/a) 2.58 ± 0.13 2.9 ± 0.15 2.87 ± 0.14 2.58 ± 0.13 2.76 ± 0.14  51 ± 4 98 ± 7 38 ± 3 37 ± 3 138 ± 10 181 ± 13 24 ± 2 280 ± 20 (n/a) 71 ± 5 24 ± 2 32 ± 2 44 ± 3 32 ± 2  320 ± 23 213 ± 16 249 ± 18 336 ± 24 236 ± 17 349 ± 24 218 ± 16 290 ± 20 (n/a) 364 ± 26 417 ± 30 426 ± 30 339 ± 24 296 ± 21  2.8 ± 0.2 2.3 ± 0.2 5.4 ± 0.4 1.5 ± 0.1 2.3 ± 0.2 1.1 ± 0.1 3.6 ± 0.3 1.3 ± 0.1 (n/a) 2.3 ± 0.2 3.1 ± 0.2 2.1 ± 0.1 2.7 ± 0.2 2.9 ± 0.2  306 ± 22 366 ± 27 259 ± 19 420 ± 30 373 ± 27 271 ± 19 336 ± 25 238 ± 17 (n/a) 299 ± 21 349 ± 25 344 ± 24 266 ± 19 224 ± 16  1940 ± 140 2170 ± 160 2170 ± 160 1930 ± 140 2040 ± 150 1660 ± 120 2270 ± 170 1680 ± 120 (n/a) 1800 ± 130 1720 ± 120 1670 ± 120 1900 ± 140 2030 ± 150  5.9 ± 0.3 7 ± 0.3 9 ± 0.4 75 ± 12 3 ± 0.1 4 ± 0.2 30 ± 1.6 2.8 ± 0.1 (n/a) 10.8 ± 0.5 7.4 ± 0.4 18.7 ± 0.9 8.9 ± 0.4 2.4 ± 0.1  4.8 ± 0.3 5.3 ± 0.4 6.9 ± 0.5 56 ± 4 2.2 ± 0.2 2.7 ± 0.4 22 ± 1 2 ± 0.1 (n/a) 8.4 ± 0.6 5.6 ± 0.4 14 ± 1 7.3 ± 0.5 1.7 ± 0.1  0 ± 0.002 0 ± 0.003 0 ± 0.004 0 ± 0.004 0 ± 0.006 0±0 0 ± 0.005 0 ± 0.005 (n/a) 0 ± 0.005 0 ± 0.005 0 ± 0.007 0 ± 0.007 0.102 ± 0.005  22 ± 1 14 ± 0.7 17.8 ± 0.9 143 ± 7 3.6 ± 0.2 4.6 ± 0.6 32 ± 1 3.7 ± 0.2 (n/a) 16.2 ± 0.8 9.9 ± 0.5 18.4 ± 0.9 9.9 ± 0.5 2.5 ± 0.2  COA = average concentration of organic aerosol in the dilution tunnel during sampling New vehicle (V14) (n/a) = data not available c Although 42 vehicles were recruited in total, vehicle #24 stalled during test #34, so the data for that test were not used). b  159  Table C.5. 100 year global warming potential (GWP100) Emission  GWP100 (confidence interval)  Source  CO2  1  IPCC, 2007  CH4  25  IPCC, 2007  CO a  1.9 (1.0 to 3.0)  IPCC, 2007  a,b  3.4 (1.7 to 6.8)  IPCC, 2007  PM: OC  a  -35 (-9 to -83)  Reynolds and Kandlikar, 2008  PM: EC  a  455 (193 to 716)  Reynolds and Kandlikar, 2008  NMHC  a  These emissions are not included in the Kyoto Protocol. NMHC = non-methane hydrocarbons Sources: Forster et al. (2007); Reynolds and Kandlikar (2008) b  Table C.6. Auto-rickshaw fuel consumption and fuel-based emission factors (mean and 95% CI) of gaseous air pollutants. Vehicles are grouped by engine type (4-stroke or 2-stroke), fuel (CNG or gasoline/petrol), and age (‘new’: 2007-2009, or ‘old’: 1998-2001) Engine, Fuel CNG-4S  PET-4S  CNG-2S  Age group (sample size) All (N=17) New (N=9) Old (N=8) All (N=11) New (N=7) Old (N=4) Old (N=13)  CO (g kg-1)  THC (g kg-1)  NOX (g kg-1)  CH4 (g kg-1)  CO2 (g kg-1)  83 (54-116) 75 (48-106) 93 (37-151) 823 (630-1013) 680 (458-863) 1073 (817-1258) 81 (47-125)  57 (40-77) 75 (49-98) 38 (19-59) 158 (97-217) 156 (96-213) 161 (34-288) 312 (276-348)  22.2 (18.2-26.3) 26 (21.6-31) 18 (12.4-23.8) 9.3 (4.7-14.1) 12.7 (7.3-18.3) 3.4 (0.3-7.3) 2.6 (2-3.2)  50 (38-62) 62 (46-78) 36 (21-52) 10 (8-12) 10 (6-13) 11 (8-14) 312 (281-344)  2630 (2560-2690) 2600 (2510-2680) 2660 (2550-2760) 1580 (1260-1980) 1820 (1410-2300) 1170 (890-1420) 1920 (1820-2030)  160  Table C.7. Fuel-based emission factors (mean and 95% CI) for fine particulate matter (PM2.5), organic carbon (OC), elemental carbon (EC) and GWC. Vehicles are grouped by engine type (4-stroke or 2-stroke), fuel (CNG or gasoline/petrol), and age (‘new’: 2007-2009, or ‘old’: 1998-2001) Engine, Fuel CNG-4S  PET-4S  CNG-2S N=11 b N=7 a  Age group (sample size) All (N=16) New (N=8) Old (N=8) All (N=10) New (N=6) Old (N=4) Old (N=13)  Fuel Cons. (kg 100km-1) 2.1 (2-2.3) 2.1 (1.9-2.2) 2.2 (2.1-2.4) 3.8 (3.4-4.2) 3.5 (3.1-3.9) 4.3 (3.6-4.7) 2.5 (2.4-2.7)  PM2.5 (g kg-1)  OC (g kg-1)  EC (g kg-1)  GWC-Kyoto (g kg-1)  GWC-All (g kg-1)  0.5 (0.3-0.9) 0.6 (0.2-1.2) 0.5 (0.2-0.7) 1.2 (0.8-1.7) a 1 (0.5-1.8) b 1.6 (1-2) 14.2 (6.2-26.7)  0.3 (0.1-0.5) 0.3 (0.1-0.7) 0.3 (0.1-0.4) 0.6 (0.3-1) 0.5 (0.2-1) 0.7 (0.4-1) 10.7 (4.8-19.1)  0.1 (0.1-0.1) 0.1 (0-0.2) 0.1 (0-0.1) 0.4 (0.2-0.5) 0.3 (0.2-0.5) 0.4 (0.2-0.7) 0 (0-0)  3880 (3600-4160) 4150 (3800-4510) 3570 (3220-3950) 1840 (1560-2190) a 2070 (1690-2490) b 1440 (1230-1650) 9710 (8960-10490)  4140 (3840-4460) 4480 (4080-4800) 3800 (3450-4160) 4060 (3880-4190) 4000 (3750-4170) 4160 (4020-4290) 9340 (8710-9970)  161  Table C.8. Comparison of distance-based emission factors from this study against other studies.  This study ARAI, 2007  Number of vehicles (and tests) CNG-4S N=17 N=1 (2) a  This study ARAI, 2007  PET-4S N=11 N=2 (4) a  2000-2009 Post 2000  3.8 ± 0.72 (n/a)  32.68 ± 16 2.13 ± 0.23  6.44 ± 5.2 0.81 ± 0.05  0.32 ± 0.25 0.47 ± 0.09  0.4 ± 0.2 (n/a)  56.8 ± 14 68.25 ± 8  48 ± 35 23 ± 10  This study ARAI, 2007  CNG-2S N=13 N=2 (4) a  1998-2001 ~2000 (retrofit)  2.5 ± 0.35 (n/a)  2.14 ± 2.3 0.69  8.12 ± 2.6 2.06  0.06 ± 0.02 0.19  7.9 ± 1.7 (n/a)  48.4 ± 3 57.7  362 ± 530 118  Kojima, 2002 b ARAI, 2007  PET-2S N=3 (43) c N=3 (6) a  1993-1996 1996-2005  4.7 ± 1.21 (n/a)  13.17 ± 6.5 1.89 ± 1.1  11.87 ± 6.8 3.4 ± 2.33  0.09 ± 0.06 0.22 ± 0.07  (n/a) (n/a)  49.3 ± 6 62.8 ± 9  670 ± 600 66 ± 40  Reference  Model year(s)  Fuel Cons. (kg 100km-1)  CO (g km-1)  THC (g km-1)  NOX (g km-1)  CH4 (g km-1)  CO2 (g km-1)  PM2.5 (mg km-1)  2000-2009 Post 2000  2.1 ± 0.25 (n/a)  1.86 ± 1.6 1.00  1.23 ± 0.8 0.26  0.46 ± 0.18 0.50  1.1 ± 0.6 (n/a)  55.9 ± 6 77.7  12 ± 14 15  LPG-2S Post 2000 ARAI, 2007 N=2 (4) a (n/a) 12.63 ± 14.9 0.88 ± 1.02 1.08 ± 0.55 (n/a) 152.0 ± 31 770 ± 600 a Emission measurements were conducted before and after maintenance for each vehicle, however the individual emission factors were not given in the report so the standard deviation is only given where more than one vehicle were tested (and their emission factors reported). b Vehicles from Dhaka, Bangladesh, (tested in Pune, India). c There were 43 tests in total because each of the 3 vehicles were tested with (a) gasoline from Dhaka vs. reference fuel, (b) with different types/amounts of lubricant, and (c) before/after maintenance (n/a) = data not available All vehicles were Bajaj auto-rickshaws tested on the Indian Drive Cycle.  162  C.4. Supporting information: References APEC. Asia Pacific Energy Consulting. India Product Specifications – Gasoline; 2010: http://www.apecconsulting.com/PDF/WebProductQualitySample.pdf. ARAI. Draft report on emission factor development for Indian Vehicles. Automotive Research Association of India, Air Quality Monitoring Project - Indian Clean Air Programme: Pune, India, 2007: http://www.cpcb.nic.in/DRAFTREPORT-on-efdiv.pdf. Bajaj. Bajaj Auto Limited, Commercial Vehicles; 2010: http://www.bajajauto.com/commercial_vehicle.asp. CPCB. Auto Fuel Quality; 2010: http://www.cpcb.nic.in/Auto_Fuel_Quality.php. DieselNet. Emission Standards, India: On-Road Vehicles and Engines; 2010: http://www.dieselnet.com/standards/in/. DOT. Pollution Under Control (PUC) Norms; Department of Transport, India, 2010: http://gov.ua.nic.in/transport/pollution_under_control.htm. EPCA. Automotive CNG fuel specifications proposed by the committee constituted by EPCA. Environment Pollution (Prevezntion & Control) Authority: Delhi, India, 2007, Report No. 29: http://www.indiaenvironmentprtal.org.in/files/Automotive%20CoNG%20fuel%20specifications. doc. Faiz, A.; Ale, B. B.; Nagarkoti, R. K. The role of inspection and maintenance in controlling vehicular emissions in Kathmandu Valley, Nepal. Atmos. Environ. 2006, 40 (31) 5967-5975. Forster, P., Ramaswamy, V.; Artaxo, P.; Berntsen, T.; Betts, R. et al. Changes in atmospheric constituents and in radiative forcing, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon, S. et al., Eds. Cambridge University Press: Cambridge, United Kingdom and New York, 2007. Gingrich, J. W.; Callahan, T. J.; Dodge, L. G. Humidity and temperature correction factors for NOX emissions from spark ignited engines. ENVIRON International Corporation: Novato, CA, 2003: http://files.harc.edu/Projects/AirQuality/Projects/H008B.2003/TH/H8BTHGasolineFinalReport. pdf. Hausker, K. Vehicle inspection and maintenance programs: International experience and best practices. US Agency for International Development, Office of Energy and Information Technology: Washington, DC, 2004: http://pdf.usaid.gov/pdf_docs/PNADB317.pdf. Kirchstetter, T. W.; Harley, R. A.; Kreisberg, N. M.; Stolzenburg, M. R.; Hering, S. V. On-road measurement of fine particle and nitrogen oxide emissions from light- and heavy-duty motor vehicles. Atmos. Environ. 1999, 33 (18), 2955-2968. 163  Kojima, M.; Bacon, R. W.; Shah, J.; Mainkar, M. S.; Chaudhari, M. K.; Bhanot, B.; Iyer, N. V.; Smith, A.; Atkinson, W. D. Measurement of mass emissions from in-use two-stroke engine three-wheelers in South Asia. SAE Tech. Pap. 2002, 2002-01-1681. Mazzoleni, C.; Moosmuller, H.; Kuhns, H. D.; Keislar, R. E.; Barber, P. W.; Nikolic, D.; Nussbaum, N. J.; Watson, J. G. Correlation between automotive CO, HC, NO, and PM emission factors from on-road remote sensing: Implications for inspection and maintenance programs. Transport. Res. D-Tr. E. 2004, 9 (6), 477-496. NIOSH. Manual of Analytical Methods. NIOSH Publication Number 2003-154 (3rd Supplement), 2003: http://www.cdc.gov/niosh/docs/2003-154/. Reynolds, C. C. O.; Kandlikar, M. Climate impacts of air quality policy: Switching to a natural gas-fueled public transportation system in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860-5865. Robinson, A. L.; Grieshop, A. P.; Donahue, N. M.; Hunt, S. W. Updating our conceptual model for fine particle mass emissions from combustion systems. J. Air Waste Manag. Assoc. (In Press). Rogers, J. Assessment of the pollution under control program in India and recommendations for improvement. World Bank, South Asia Urban Air Quality Management Program: Washington, DC, 2002: http://siteresources.worldbank.org/PAKISTANEXTN/Resources/UrbanAir/MainReport.pdf. Subramanian, R.; Khlystov, A. Y.; Cabada, J. C.; Robinson, A. L. Positive and negative artifacts in particulate organic carbon measurements with denuded and undenuded sampler configurations. Aerosol. Sci. Tech. 2004, 38 (S1), 27-48. Volckens J.; Braddock, J.; Snow, R. F.; Crews, W. Emissions profile from new and in-use handheld, two-stroke engines. Atmos. Environ. 2007, 41 (3), 640-649.  164  D. Appendix D: Supporting information for Chapter 5 Table D.1. Distance-based emission factors for auto-rickshaws with 2-stroke engines: base datasets and policy-impacted groups (uncertainty shown by median and inter-quartile range). Vehicles/Policies a (n = # of vehicles in sample after policy implementation) PET-2S: Base Data (n=46) I/M: Idle Visible PM (n=36) I/M: Dyno Gaseous (n=25) I/M: Dyno PM (n=11)  Fuel Cons. (kg/100km)  CO2 (g/km)  CO (g/km)  CH4 (g/km)  NMHC (g/km)  NOX (g/km)  OC (mg/km)  EC (mg/km)  PM2.5 (mg/km)  3 (2.9, 4.2) 3 (2.9, 3.1) 2.9 (2.7, 3) 3.1 (2.9, 3.3)  49 (46, 52) 50 (47, 53) 50 (48, 56) 50 (49, 58)  13 (7, 16) 11 (6, 14) 7 (4, 11) 10 (6, 15)  0.42 (0.37, 0.78) 0.4 (0.36, 0.45) 0.39 (0.36, 0.44) 0.38 (0.22, 0.46)  7.9 (7, 15) 7.6 (7, 9) 7.4 (7, 8) 7.2 (4, 9)  0.08 (0.04, 0.12) 0.1 (0.04, 0.15) 0.09 (0.06, 0.16) 0.06 (0.04, 0.18)  308 (200, 650) 244 (180, 350) 240 (180, 340) 143 (100, 170)  21 (13, 43) 16 (12, 24) 16 (12, 23) 10 (7, 11)  410 (270, 870) 325 (240, 470) 320 (240, 450) 190 (140, 230)  CNG-2S: Base Data 2.6 48 1.1 7.7 1.6 0.06 142 184 0 (0, 0) (n=13) (2.2, 2.8) (46, 50) (0.9, 2.1) (6.9, 8.3) (0, 2) (0.05, 0.07) (70, 220) (110, 280) I/M: Idle Visible PM 2.6 46 1.8 7.5 1.6 0.06 74 109 0 (0, 0) (n=7) (2.4, 2.8) (45, 50) (1.1, 4) (6.9, 7.7) (0.2, 1.8) (0.05, 0.07) (50, 170) (80, 210) I/M: Dyno Gaseous 2.1 47 1.3 7.5 0.06 120 150 n/a 0 (0, 0) (n=4) (2, 2.2) (45, 50) (1, 2.6) (6.7, 7.8) (0.05, 0.07) (70, 200) (100, 260) I/M: Dyno PM 2.6 49 1.3 7.5 1.6 0.07 110 150 0 (0, 0) (n=9) (2.2, 2.8) (46, 50) (0.9, 3.1) (6.9, 7.9) (0, 2) (0.05, 0.08) (60, 140) (90, 180) a Primary data for CNG-2S and CNG-4S from Indian Auto-Rickshaw Project (Reynolds et al. 2009); primary data for PET-2S and PET-4S from Kojima et al. (2002) and ARAI (2007). b EC measurements for 2-stroke vehicles were below limits of detection due to large fraction of OC. n/a = data not available  165  Table D.2. Distance-based emission factors for auto-rickshaws with 4-stroke engines: base datasets and policy-impacted groups (uncertainty shown by median and inter-quartile range). Vehicles/Policies a (n = # of vehicles in sample after policy implementation) PET-4S: Base Data (n=13) New Vehicle (n=1) Scrap Old Vehicles (n=9) I/M: Idle Gaseous (n=6) I/M: Dyno Gaseous (n=3) I/M: Dyno PM (n=9)  Fuel Cons. (kg/100km)  CO2 (g/km)  CO (g/km)  CH4 (g/km)  NMHC (g/km)  NOX (g/km)  OC (mg/km)  EC (mg/km)  PM2.5 (mg/km)  3.9 (3.3, 4.3)  59 (47, 63)  27 (22, 36)  0.36 (0.3, 0.5)  2.3 (1, 9)  0.4 (0.1, 0.5)  19 (9, 32)  12 (9, 21)  30 (23, 62)  2.3  72  1.8  0.01  1.4  0.6  0.1  0.05  0.4  3.6 (3.1, 4) 3.6 (3.3, 4)  61 (58, 72) 67 (53, 73) 72 (67, 73) 59 (50, 72)  23 (2, 27) 12 (2, 32) 1.97 (1.88, 2.13) 23 (2, 35)  0.33 (0, 0.5) 0.2 (0, 0.4) 0.04 (0.02, 0.04) 0.36 (0, 0.4)  2.3 (1, 6) 1.1 (1, 2) 0.8 (0.8, 1.1) 1.6 (1, 6)  0.5 (0.4, 0.5) 0.5 (0.2, 0.6) 0.53 (0.47, 0.59) 0.4 (0.1, 0.5)  10 (8, 23) 8 (4, 19) 8 (4, 11) 9 (7, 10)  12 (9, 19) 8 (4, 9) 5 (3, 8) 9 (8, 10)  24 (21, 33) 24 (16, 29) 15 (8, 23) 24 (21, 30)  1.1 (0.5, 1.5)  0.1 (0, 0.3)  0.4 (0.3, 0.6)  4.4 (0.6, 8.3)  1.7 (0.2, 3.1)  10 (1.1, 17.1)  1.7  0.07  0.4  0.32  0.11  0.41  1.5 (1, 1.7) 1 (0.4, 1.5)  0.2 (0, 0.6) 0.1 (0, 0.2)  0.5 (0.4, 0.6) 0.5 (0.3, 0.6)  3.2 (0.4, 8.9) 3.7 (0.6, 8.3)  1.7 (0.2, 3.2) 1.6 (0.2, 3.2)  5.9 (0.8, 17.3) 8.9 (1.1, 17.1)  2.3 3.6 (3.3, 4)  CNG-4S: Base Data 2.1 55 1.3 (n=17) (2, 2.2) (52, 57) (0.6, 2.7) New Vehicle 1.8 46 0.7 (n=1) Scrap Old Vehicles 2.1 52 1.3 (n=9) (1.9, 2.2) (51, 55) (0.8, 2.3) I/M: Dyno Gaseous 2 55 0.8 (n=13) (1.9, 2.2) (52, 57) (0.6, 1.7) a Primary data from Indian Auto-Rickshaw Project (Reynolds et al. 2009)  166  Figure D.1. Assessment of emission-control policies using the health-climate framework. Errorbars include uncertainty in emission factors (both axes) as well as uncertainty in global warming potential for climate-forcing species (y-axis). A. All policies. B. Policies applied to vehicles with 4-stroke engines only, with the x-axis covering a smaller range so the impact of different policies can be distinguished. Note that the y-axes in both panels are to the same scale to facilitate comparison.  167  Figure D.2. Annual reduction in PM2.5 emissions from a fleet of 5000 auto-rickshaws (tonnes per year). A. Auto-rickshaws with 2S engines. B. Auto-rickshaws with 4S engines. Each axis within a radar plots has the same scale, to show the relative effectiveness of policies. However the axes on Panels A and B are on different scales, so 10 tonnes per year has been highlighted in red font to facilitate comparisons.  168  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0071479/manifest

Comment

Related Items