Mobile sources critical review: 1998 NARSTO assessment

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/221947440

Mobile sources critical review: 1998 NARSTO assessment ARTICLE in ATMOSPHERIC ENVIRONMENT · DECEMBER 2000 Impact Factor: 3.28 · DOI: 10.1016/S1352-2310(99)00463-X

CITATIONS

READS

159

28

6 AUTHORS, INCLUDING: Robert A Harley University of California, Berkeley 150 PUBLICATIONS 5,600 CITATIONS SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.

Available from: Robert Sawyer Retrieved on: 05 February 2016

MOBILE SOURCES CRITICAL REVIEW 1998 NARSTO ASSESSMENT

Submitted to ATMOSPHERIC ENVIRONMENT

R.F. Sawyer, a R.A. Harley,b S.H. Cadle,c J.M. Norbeck,d R. Slott,e H.A. Bravof

17 June 1998

Key Words: motor vehicle emissions, emissions inventory, mobile sources, in-use emissions, emissions uncertainties

a

University of California at Berkeley, Department of Mechanical Engineering, Berkeley CA 94720-1704 USA; email: [email protected] (to whom correspondence regarding this paper should be directed) b University of California at Berkeley, Department of Civil and Environmental Engineering, Berkeley CA 97420-1710 USA; email: [email protected] c General Motors R&D, Health and Environment Department, MC480-106-269, 30500 Mound Road, Warren MI 48090-9055 USA; [email protected] d University of California at Riverside, CE-CERT, Riverside CA 92521; email: [email protected] e Consultant, 71 Hawes Avenue, Hyannis MA 02601 USA; email: [email protected] f Universidad Nacional Autonoma de Mexico, Centro de Ciencias de la Atmosfera, Seccion de Contaminacion Ambiental, Circuito Exterior, Ciudad Universitaria, Mexico, D.F. 04510, MEXICO; email: [email protected]

ABSTRACT Mobile sources of air pollutants encompass a range of vehicle, engine, and fuel combinations. They emit both of the photochemical ozone precursors, hydrocarbons and oxides of nitrogen. The most important source of hydrocarbons and oxides of nitrogen are light- and heavy-duty onroad vehicles and heavy-duty off-road vehicles, utilizing spark and compression ignition engines burning gasoline and diesel respectively. Fuel consumption data provide a convenient starting point for assessing current and future emissions.

Modern light-duty, gasoline vehicles when new have very low emissions. The in-use fleet, due largely to emissions from a small “high emitter” fraction, has significantly larger emissions. Hydrocarbons and carbon monoxide are higher than reported in current inventories. Other gasoline powered mobile sources (motorcycles, recreational vehicles, lawn, garden, and utility equipment, and light aircraft) have high emissions on a per quantity of fuel consumed basis, but their contribution to total emissions is small. Additional uncertainties in spatial and temporal distribution of emissions exist.

Heavy-duty diesel vehicles are becoming the dominant mobile source of oxides of nitrogen. Oxides of nitrogen emissions may be greater than reported in current inventories, but the evidence for this is mixed. Oxides of nitrogen emissions on a fuel-consumed basis are much greater from diesel mobile sources than from gasoline mobile sources. This is largely the result of stringent control of gasoline vehicle emissions and a lesser (heavy-duty trucks) or no control (construction equipment, locomotives, ships) of heavy-duty mobile sources.

The use of natural gas, propane, alcohols, and oxygenates in motor vehicles is increasing but remains small. Vehicles utilizing these fuels can be but are not necessarily cleaner than their gasoline or diesel counterparts. These alternative fuels represent a small fraction of total fuel consumption and of total pollutants emitted.

Historical vehicle kilometers traveled growth rates of about 2% annually in both the United States and Canada will slow somewhat to about 1.5%. Mexican growth rates are expected to be greater. Fuel consumption growth in recent years of about 1.4% annually is projected to continue as slowing gains in fuel economy from fleet turnover are more than offset by growth. This growth also will erode the emissions reductions resulting from cleaner new vehicles and fuels. Uncertainties in these projections are high and affected by economic activity, demographics, and the effectiveness of emissions control programs—especially those for reducing in-use emissions.

RATIONALE AND OVERVIEW Mobile source emissions come from a range of vehicle, engine, and fuel combinations, listed in Table 1. Those most important to tropospheric ozone, as measured by their hydrocarbon (HC) and nitrogen oxides (NOx) emissions, and of lesser importance, their carbon monoxide (CO) emissions, are shown in bold. This review emphasizes these most important categories with a briefer consideration of the other mobile sources. Historically, the dominant mobile source emissions have been those from on-road vehicles, powered with reciprocating internal combustion engines burning gasoline or diesel fuels. As these sources have come under control, the importance of largely uncontrolled emissions from off-road motor vehicles, aircraft, ships, and locomotives has increased.

FUEL BASED EMISSIONS A common approach to estimating road vehicle emissions is to combine estimated average emissions per distance traveled, grouping classes of vehicles, with activity patterns of distance traveled, preferably with temporal and spatial resolution. The emissions estimates derive from direct measurements (light-duty vehicles on dynamometers), indirect measurements (heavy-duty engines on dynamometers combined with vehicle loads), tunnel measurements, and models which combine experimental measurement and physical approximations. Two widely used emissions models are USEPA’s MOBILE5B and the CARB’s MVEI. These models are reviewed in a later section. This approach can be extrapolated in some cases to estimate emissions from other transportation modes.

Another approach, which is applicable across all modes, is a fuel based emissions model. Emissions are expressed as an “emissions index” (gm pollutant/kg fuel), emission rates determined by combining the emission index with the fuel consumption rate, and total emissions from the product of the emissions index and total fuel consumption. The fuel-based approach allows intermodal comparisons. It also provides some advantages in estimating road vehicle emissions which are often approximately proportional to fuel consumption over a range of operating conditions. It also provides a means of approaching mobile source emissions in Mexico for which the vehicle fleet and its emissions are not as well characterized as in the United States and Canada. Fuel consumption data are usually known with greater precision than vehicle usage. Of course, much of the same measured emissions data and many of the same assumptions are common to both approaches.

TRANSPORTATION FUELS Gasoline and diesel are the dominant on-road and off-road fuels. Both have “reformulated” forms (RFG and RFD) which reduce emissions and are in use in regions of the United States having the most severe air pollution. Additional areas will use these fuels in the future and additional reformulation will occur with the introduction of the USEPA’s Phase II RFG. While national and local policies encourage the use of alternative fuels (natural gas, liquid petroleum gas, methanol, ethanol, and even hydrogen and electricity), they are a minor factor in emissions. The largest use of “alternative” or non-petroleum-based fuels is in the form of oxygenates added to gasoline. Emissions associated with fuel processing appear in the stationary source inventory, but are related to transportation fuel usage. TIME FRAMES

Our review focuses on the current understanding of mobile source emissions estimation, especially the uncertainties. The year 2010 is a date of future interest. It is the current deadline for the most severe air pollution region (the Los Angeles Air Basin) to come into ozone compliance and it is the approximate time for meeting the new ozone air quality standard. Both fuel usage and the nature of the vehicle fleet affect future emissions. For the past ten years, fuel consumption has been increasing by about 1.4% per year (EIA, 1994), caused by increasing vehicle kilometers traveled, the growth of the heavy-duty fleet, and a shift from personal passenger cars to personal light-duty trucks, vans, and sport utility vehicles. This increase in fuel consumption is expected to continue. In general, the vehicle fleet is getting cleaner as new, lower emitting vehicles with increased durability replace older vehicles. The change in mobile source emissions involves the trade-off in the increase in fuel consumption and the decrease in the in-use

emissions of the vehicle fleet. The uncertainties of the latter are the greater. Future emissions will depend primarily upon lifetime emissions from vehicles, which are now more strongly affected by deterioration, maintenance, and the effectiveness of inspection and repair programs than by new car emissions standards. Estimates of emissions from the both the current and future fleets contain large uncertainties. FUEL CONSUMPTION DATA The inventory of transportation fuel consumption (excluding that for pipelines) provides both a good starting point for mobile source emissions estimation, and a basis for extrapolation into the future. “Transportation” and “mobile sources” are largely, but not entirely the same. Pipelines are transportation but their associated emissions are assigned to stationary sources. Garden equipment, some off-road recreational vehicles, railroad refrigeration engines, and other sources may not be considered transportation but generally are included in the mobile source inventory.

Figure 1 summarizes fuel consumption in the mobile source sector for North America, as of the mid-1990s. Data for Canada (StatCan, 1997) and the United States(ORNL, 1996) are readily available. Data on mobile source fuel consumption in Mexico(PetMex, 1997) are limited. Overall, the United States accounts for 87% of North American fuel consumption by mobile sources. Use of fuel by mobile sources in much less in Canada and Mexico: these countries account for 7 and 6%, respectively, of total North American mobile source sector fuel use. MOTOR VEHICLE FLEETS The approximately 230 million road vehicles in North America are about 68% passenger cars, 26% light-duty trucks, and 6% heavy-duty trucks and buses, Table 2. The average age of

passenger cars in the United States is 8.6 years, an increase of one year over the past ten years(AAMA, 1997). Thirteen percent of these passenger cars are more than 15 years old. This aging of the fleet probably results from a combination of increased vehicle durability and increased new vehicle cost. The average truck is 8.3 years with 19% more than 15 years old. Compared with passenger cars, the truck fleet contains a higher fraction of both young (because of recent increase in sales of light-duty trucks) and old (because of the longer service life of commercial trucks) vehicles. Vehicle registration in the United States over the past ten years has grown at an annual rate of about 2%, dominated by a 4.5% annual growth in trucks. Mexico vehicle registrations have increased at an annual rate of 5.5% over the past ten years (IMT, 1994, INEG, 1997). The United States fleet includes more than 350,000 alternative fueled vehicles (less than 0.2%), the majority of which operate on propane (EIA, 1996). The Mexican vehicle population is concentrated in highly populated and industrialized areas of Mexico. Mexico City, Monterey, Guadalajara, and the Northern Border of Mexico Zone contain nearly half of Mexico’s vehicles.

GASOLINE ENGINES AND MOBILE APPLICATIONS Gasoline is the dominant fuel of light-duty road vehicles, medium-duty road vehicles, recreational vehicles, garden equipment, and the general aviation fleet. This fuel-engine combination provides a high energy-density, low-cost engine that probably explains its dominance in these applications. Light-duty vehicles Light-duty gasoline vehicles use several systems to control emissions. The positive crankcase ventilation (PCV) system was the first control placed on vehicles. It directs blowby gases back to

the air inlet for reprocessing in the engine. The exhaust gas recirculation (EGR) system directs a small fraction of the exhaust back into the engine to dilute the intake charge and reduce peak combustion temperature, hence reducing the engine-out NOx. Exhaust gases are treated by the three-way catalyst system. The term “three-way” refers to the catalyst’s ability to oxidize CO and HC to CO 2 while simultaneously reducing NOx to N 2. Proper function requires that the engine operate at the stoichiometric air-to-fuel ratio, the point at which exactly enough air is ingested to oxidize all of the fuel. An oxygen sensor in the exhaust system continuously monitors the air-to-fuel ratio in the exhaust system. Both the catalyst and oxygen sensor must be hot to operate. Thus, they are not functional during cold start operation. Minimizing the resulting cold start emissions is a major focus of current control efforts. The primary method has been the development of catalysts that can be mounted close to the engine. These catalysts must be able to withstand extreme temperature excursions.

The introduction of computer controlled fuel injection has greatly improved the ability to maintain stoichiometric operation. Fuel injection on most current production vehicles is done in the inlet air system, with the more sophisticated systems using multiple injectors timed to the cylinder intake cycle (sequential multi-port fuel injection). Development efforts are currently focusing on direct injection engines. The engine control module (an on-board computer) monitors and controls the engine operation. Current algorithms recognize when engine operating parameters must be changed to optimize emissions performance and store the new parameters for the future. This process is referred to as adaptive learning. New vehicles are now using the second-generation on-board diagnostics system (OBDII) to monitor emissions control system performance. A combination of sensors and algorithms are used by this system to monitor the performance of the

engine, the catalyst, and the evaporative emissions control system. If a problem is detected the check engine light is turned on and an error code is stored for diagnostic repair purposes. For some emission system failures engine operation is changed to a “limp-home” operation mode that reduces emissions.

The evaporative emission control system is designed to capture HC emissions that occur both during operation and while the vehicle is parked. HC vapors from the fuel are routed to a charcoal canister where they are stored while the vehicle is parked. The canister is rejuvenated via air purging during vehicle operation. Evaporative emissions also are caused by very small fuel leaks and permeation through hoses and plastic fuel tanks. Current efforts are focused on improved materials and fittings to ensure good long-term in-use performance. In addition, a recent regulation requiring that refueling vapors be captured on-board has resulted in the addition of the on-board refueling vapor recovery (ORVR) system as well.

Finally, it is best to view the vehicles as a vehicle/fuel system, since the fuel impacts both the exhaust and evaporative emissions. Leaded fuel has been completely phased out in the US and Canada, and is being phased out in Mexico, since lead is toxic and a catalyst poison. More recently, seasonal fuel vapor pressure limits have been mandated. These significantly lower evaporative emissions. Some areas with wintertime exceedances of the CO air quality standards require the addition of oxygenates to the fuel in the winter. This provides benefits for older vehicles by essentially ‘leaning’ the mixture. However, for modern vehicles, with full closed-loop control and adaptive learning, there is little or no benefit - in terms of reduced emissions - arising from the addition of oxygenates. Finally, reformulated gasolines have been put into use in all of

California and in severe ozone nonattainment areas. One of the features of the California reformulated gasoline is a mandated low sulfur content. Sulfur is a mild catalyst poison, whose effects tend to be reversible. A variety of alternative fuels (e.g., natural gas, 85% alcohol/15% gasoline blends, and propane) can be used in vehicles designed for their use. However, the lack of an infrastructure for these fuels together with cost considerations, and the improvements in the emission reductions from new vehicles and gasolines, have severely restricted interest in the use of these alternative fuels. Medium- and heavy-duty vehicles Emission control technologies for medium- and heavy-duty applications are similar to those used on light-duty vehicles, but their introduction has lagged somewhat. PCV systems were introduced in 1968, evaporative canisters and EGR systems in 1985, and the widespread use of catalysts in 1987. Less stringent oxides of nitrogen standards, compared with light-duty vehicles, allow the continued use of oxidation only catalysts on many vehicles. Other gasoline engine applications Other applications of gasoline engines include on-road motorcycles, recreational land vehicles, recreational watercraft, lawn, garden and utility equipment, and light aircraft and general aviation. Emissions per kilogram of fuel are high because these applications are either unregulated or only recently regulated. Because fuel consumption in these sectors is low, contributions to total emissions are small compared to on-road emissions from light-, medium-, and heavy-duty road vehicles but will be of increasing importance as the on road sources are controlled.

FUELS AND FUEL USE

Gasoline Commercial gasoline is a complex mixture of many hydrocarbons and oxygen-containing compounds that are blended to provide combustion characteristics compatible with the engines in which the gasoline is burned. The most important characteristics affecting combustion are vapor pressure and octane number. These and other parameters, such as aromatic, olefin, sulfur, and additive content and distillation temperatures affect emissions. In Canada and the United States nearly all gasoline consumed is unleaded, the exception being aviation gasoline. Unleaded, highoctane aviation gasolines are being studied but there is no schedule for phase-out of the current fuel. In Mexico, 93% of the gasoline sold is unleaded (IMP, 1997). Reformulated gasoline The modification of the composition of gasoline to reduce emissions has resulted in a series of “reformulated” gasolines. The addition of an oxygenate in winter reformulated gasoline reduced carbon monoxide levels. Lower vapor pressure, reduced aromatic and olefin content, addition of oxygenate, reduced sulfur, and a narrowed distillation range (lower T90 and T50 and higher T10) provide emissions reductions (AOAQIRP, 1997). Such reformulated gasolines have been introduced in Canada, the United States, and Mexico as ozone control strategies. The effectiveness can vary depending upon local HC/NOx ratios. Reformulated gasolines tend to be more effective at reducing hydrocarbons than NOx. The most severely reformulated gasoline is California Phase II, which typically has very low sulfur (31 ppm), a low Reid vapor pressure (51 kPa), a narrow distillation range (T10=51oC, T 90=145oC), and reduced aromatics (25%), olefins (4%), and benzene (0.9%), and added MTBE (11%).

EMISSIONS

Regulations and test methods

Light-duty vehicles National emission standards for vehicles were first promulgated in the Clean Air Act published in the Federal Register in 1966. These standards applied to the 1968 model year and were similar to emission standards set in California in 1960. However, the 1970 amendments to the Clean Air Act established the initial comprehensive motor vehicle emissions standards that required a 90% reduction of tailpipe emissions from the then uncontrolled vehicles. These standards applied to 1975 vehicles and initially covered only carbon monoxide and hydrocarbon emissions. This was later expanded in 1977 to include oxides of nitrogen and particulate matter that allowed for diesel vehicles to be included in the regulations also. These Federal standards applied to the 49 States excluding California. California was given the authority to establish their own more stringent standards because of the severity of the air pollution in the Los Angeles area. This situation still exists today although the Federal and California standards and the associated emission control technology have merged including several states adopting the low emission vehicle standards recently established in California. Tailpipe emissions The evolution of the United States light-duty vehicle emission standards is shown in Figure 2. The regulations are given in terms of mass of pollutant/distance (g/mi) and are the accumulated emissions over the Urban Dynamometer Driving Schedule (UDDS) or Federal Test Procedure (FTP) after 80,000 or 160,000 kilometers. Oxides of nitrogen were first controlled in 1976; the increase in NOx emissions in 1968 was the result of modifications to control CO and HC.

Crankcase emission control were first applied in the 1960s but only officially regulated in 1988. The FTP driving cycle is shown in Figure 3. This cycle consists of three phases with a total cycle time of 1877 seconds and an additional 600 seconds hot soak time. The first cycle is 505 seconds, which includes a cold engine start after soaking the vehicle overnight at an ambient temperature of 18 oC. The second cycle is 867 seconds of stabilized driving. Following the 10-minute soak with the engine off, another 505 seconds transitory phase is driven which includes a hot start. The emissions of each phase are weighted differently (0.43 for the cold phase; 1.00 for the stabilized phase; and 0.57 for the warm phase) to account for driving and trip characteristics.

The FTP driving cycle was adopted as the emissions certification test cycle in the early 1970s. It has been widely used for emissions inventory purposes as well, although it does not fully represent modern driving behavior that includes higher speeds and more aggressive driving. Most manufacturers use an enrichment strategy to improve power and to protect catalysts from overheating at high power operation. This greatly increases CO and HC emissions, but only at driving conditions that are outside of the FTP speed and acceleration envelope. In response, both the USEPA and the CARB have established driving cycles that are supplementary to the FTP. These include the US06 driving cycle, which emphasizes high speed and load to regulate the time in enrichment and the SC03 cycle to test the impact of air conditioning load. Other driving cycles for light-duty vehicles include the Highway Fuel Economy test (HWFET), the New York City Cycle, and the California Unified Cycle (LA-92). A number of cycles have been developed to better test for emissions representative of in-use driving. The Unified Cycle is a popular example of a cycle developed for emission inventory purposes rather than vehicle certification.

All tested passenger vehicles are driven on a chassis dynamometer that simulates the load and inertia of the vehicle on the road. The test fuel is indolene, a special, low sulfur, tight specification gasoline. The exhaust gases are diluted with filtered ambient air to maintain a constant total flow of exhaust and air during all driving conditions. This ensures a constant volume sample (CVS) for measurement of both gaseous and particulate emissions. For particulate emissions a dilution tunnel is added to provide controlled conditions for the collection of the particulate matter on filters for mass determination. Light-duty trucks In the United States separate emission standards for light-duty trucks (less than 3860 kg GVW) were promulgated in 1975 and, similarly to passenger vehicles, have been continually revised downward. Nearly identical standards were implemented for both Federal and California trucks in 1993-94. In addition, California has established low emitting truck standards that are to be phased in between 1994 and 2003. There are now three separate truck standards based on vehicle weight. The test methods used for passenger vehicles are also used for the three truck classes.

Mexico and Canada also use the US FTP for their emissions certification. Canadian emission standards for light-duty vehicle were less stringent than US standards from 1975 through 1987 but in 1988 became similar to those in the US. Mexico implemented 1981 US Federal emission standards in 1993 and is moving toward standards similar to US Tier I. Evaporative emissions In addition to tailpipe emissions, the US has established emission standards for evaporative emissions. Evaporative emissions are divided into five types:

•

Diurnal, which are the emissions when the vehicle is at rest which occur due to ambient temperature changes over a typical 24-hour period (the portion of emissions at rest driven by the impact of temperature on the vapor pressure above the fuel).

•

Hot Soak, which is driven by residual engine heat once a warmed-up vehicle is parked and the engine is shut off.

•

Running Loss, which occur when the vehicle is being driven.

•

Resting Loss (the constant at rest evaporative emissions).

•

Refueling Loss (displaced vapors and drippage resulting from refueling).

The Sealed Housing Evaporative Determination (SHED) method is used throughout North America to measure evaporative emissions. The vehicle resides in the SHED during the tests. Emitted hydrocarbons are measured to determine the evaporative emission rate. Procedures for measuring evaporative emissions were revised by USEPA in 1993 and consist of vehicle preconditioning followed by exhaust emission testing, a hot soak test, a running loss test, and a diurnal emissions test. Earlier test methods, which were used to generate much of the data used in emissions inventories, have been recognized as inadequate. A full test now takes five days and involves higher temperatures and greater temperature changes than the original procedure. The CARB has developed a similar test, as well as an alternative procedure. Medium- and heavy-duty vehicles Unlike passenger vehicles and light trucks which are certified and tested as vehicle families, heavy-duty diesel engines are certified and tested on an engine dynamometer, over a transient test procedure, which includes both cold start and warm start operation. The transient test procedure

was developed from engine data collected in New York and Los Angeles, during freeway and nonfreeway operation. The transient test procedure consists of a prescribed engine speed and load (rpm and torque) schedule, which is specific to each engine, and is developed from the maximum torque curve (or maximum torque versus engine speed) for the engine. Cold start and warm start operation are weighted 1/7, and 6/7, respectively. There is a 20-minute soak period between cold and warm operation. Emissions data and engine load data (in brake horsepower) are collected over the test procedure, and emission rates of heavy-duty engines are generally reported in grams/bhphr. Other gasoline engine emissions regulations On-road motorcycles over 50 cm3 displacement were first regulated in 1978 by both the USEPA and CARB. Only the CARB regulates emissions from off-road motorcycles. Emissions controls on some recreational vehicles were first implemented in 1996 year in California. The USEPA recently established emissions regulations for some recreational watercraft that will first apply in 1998. Both the USEPA and CARB have implemented emissions standards for lawn, garden, and utility equipment under 25 horsepower. Real world vehicle emissions Real world vehicle emissions are difficult to predict because many factors are involved. When vehicles are new, emissions per kilometer are less than or equal to Federal or California requirements. However, as vehicles age and accumulate mileage, deterioration of fuel system and emission control components can result in much higher emission rates. In some high emitting vehicles HC and CO emissions per kilometer are over two orders of magnitude higher than when those same vehicles were new. These high emitting vehicles have been found to have broken fuel

system or emission control components which, when repaired, can return the vehicle to near normal emissions levels (Stephens and Cadle, 1991).

In the future the amount of time that well-maintained vehicles give high emissions will be greatly reduced. New catalyst technologies are being developed to reduce cold start emissions. USEPA will restrict the amount of fuel rich operation a vehicle can experience with the introduction of a new high power test cycle. In addition, evaporative systems will become much more robust, also in response to a more stringent test procedure.

The rate at which high exhaust emitting vehicles occur is decreasing. Significant changes have occurred in vehicle emission control technology since the early 1980s and these changes have reduced the rate of vehicle deterioration with age and mileage accumulation (MSTAC, 1997). The technologies include port fuel injection, more durable catalysts, and more precise fuel controls. It is anticipated that On Board Diagnostic technology (OBD II) on all 1996 and later vehicles will further reduce vehicle emissions deterioration.

A series of real world evaporative emissions studies have been completed recently. SHED tests have been made on randomly selected vehicles to determine hot soak emissions (Brooks et al., 1995), diurnal emissions (Haskew, 1997a), and running loss emissions (Haskew, 1997b). Vehicles selected for these tests were solicited so that the acceptance rates were very high to minimize potential selection bias. No repairs or adjustments were made to the evaporative emissions control systems prior to the tests. Many different component failures were observed. Emissions were normally distributed among vehicles except for a few vehicles that had very high emissions,

primarily due to fuel leaks. Pierson et al. (1997) have provided a thorough review of evaporative emissions studies.

The distribution of in-use emissions among vehicles for each of exhaust HC, CO and NOx, and diurnal, hot soak, running loss HC have been found to be highly skewed. A small percent of vehicles contribute a significant amount to total fleet emissions. However, the chance of a high emitter for one pollutant being a high emitter from another is low except for exhaust HC and CO, where fuel rich conditions can cause both HC and CO to be high.

The emissions distributions of CO, HC, and NOx from motor vehicles are highly skewed, with a small number of gross-polluting vehicles, about 10% of the vehicle fleet, responsible for half of the total emissions for each pollutant. There is significant overlap in the subsets of the vehicle fleet that are high-emitters of CO and HC. The NO x high-emitters comprise a different, mostly disjoint set of vehicles from the CO and HC gross-polluters. This skewness in the distribution makes characterization of emissions by the testing of randomly selected vehicles difficult. A large number of vehicles need to be tested to ensure that the high emitters are present in statistically significant numbers. The vehicle selection process for testing must be random. Special care must be taken to avoid a selection bias that would omit high emitting vehicles from the tested sample of the population.

A number of techniques have been used in attempting to measure emissions from large numbers of real world vehicles. These include remote sensing of tailpipe exhaust, analysis of IM240 results from state inspection/maintenance programs, random roadside pullover tampering studies,

tunnel studies, and ambient speciated hydrocarbon measurements. Remote sensing, and IM240 measurements give information on individual vehicle tailpipe emissions while tunnel tests and analyses based on ambient measurements can give information about both exhaust and evaporative emissions but only on fleets of vehicles. Tunnel and ambient air studies are reviewed in later sections of this paper.

Remote sensing of thousands of vehicles at sites where the catalyst would be working if operational (i.e., under light acceleration in a hot stabilized mode) can be used to characterize tailpipe emissions in vehicle fleets measured at the site. Since vehicle license plates can be identified simultaneously, the dependence of tailpipe emissions on age and other vehicle characteristics can be determined. Emissions measured by remote sensing in Denver, Colorado had a vehicle age dependence essentially the same as that observed by IM240 measurements of a similar fleet in the same area. Since the vehicles measured by remote sensing were identified individually, and each could be classified as to whether it had taken a new I/M test, the measurements were used to estimate the effectiveness of the new I/M control strategy program (Stedman et al., 1997). Observed emission reductions were less than predicted.

The use of IM240 testing in state inspection/maintenance programs offers an excellent opportunity to monitor emissions from millions of real world vehicles. Current testing protocols allow for a “quick pass” of clean vehicles that limits the comparability of data. Variable waiting times before testing causes some vehicles to have artificially high emissions. A few states are collecting data on a random multi-thousand sample of vehicles measured on a full IM240 test.

These data would be improved if measurements were made after the vehicles were appropriately pre-conditioned (MSTAC, 1997).

Tunnel tests have the advantage that almost all the emissions measured are coming from vehicles passing through the tunnel. The hydrocarbon species present are not subject to significant photochemical degradation. The hydrocarbon speciation can be used to separately estimate the amount of tailpipe and running loss emissions. Since tailpipe, vapor loss, and liquid loss hydrocarbons contain common species, assignment of the correct source can be difficult.

Ambient measurements of hydrocarbons have been used to estimate the percent contribution of hydrocarbons from mobile sources. Key hydrocarbon species are used to construct source profiles for different emission sources. Location of the samplers can be used to further clarify emission sources. Uncertainties in these estimates are associated with location of the sampling instruments, comprehensiveness of the source profiles, accuracy of the source profiles, and reactivity of some hydrocarbon species. Inspection and maintenance An effective inspection and maintenance program can reduce in-use emissions, especially those from the high emitter fraction of the fleet. These programs are really maintenance, inspection, and repair with inspection serving three purposes, 1) promotion of better maintenance, 2) identification of high emissions and their cause leading to repair, and 3) verification of the effectiveness of repair. The effectiveness of many current programs is in question and the resulting emissions reductions uncertain (Pierson, 1996a). Effective inspection and maintenance programs, however, are an essential part of dealing with in-use emissions from the total motor

vehicle fleet, especially, but not exclusively, the oldest vehicles. Both the regulators and industry anticipate that on-board diagnostics (OBD) will improve inspection and maintenance effectiveness. In the absence of an effective program, fleet in-use emissions will remain a large multiple of new car emission standards. This area is critical to limiting emissions from the in-use highway fleet, both light- and heavy-duty. The assessment of the current or future effectiveness of inspection and maintenance programs and, therefore, the reliable projection of future emissions, is difficult.

An important secondary benefit from inspection and maintenance programs is the collection of emissions data for millions of vehicles representative of the in-use fleet. These data, especially those coming from programs that measure emissions in gm/km, can reveal the distribution of emissions among vehicles by age, model, and technology. In future years these data will provide a record of reductions in emissions associated with the introduction of both new technology and fuels.

DIESEL ENGINES, MOBILE APPLICATIONS, AND EMISSIONS REGULATIONS

Light-duty vehicles Light-duty diesel engines have seen very limited use in North America. Their primary advantage is better fuel economy than gasoline engines. However, this is not enough of an incentive to overcome the higher cost of the engines and a reputation for odor, noise, poor acceleration, and hard starting. Major advances have been made which have essentially eliminated the odor and performance problems. The modern diesel engine is now a high-speed, direct injection engine that

can be turbocharged. A number of technologies are used to reduce exhaust emissions. These include combustion chamber design, exhaust gas recirculation (EGR), and oxidation catalysts (Krieger et al., 1997). The oxidation catalyst has helped eliminate the traditional diesel odor. Future engines are expected to be high speed, direct injection engines that use high-pressure electronically controlled fuel injection and high levels of EGR. Despite these advances, the diesel continues to have trouble meeting increasingly stringent exhaust emissions standards for NOx and particulate matter (PM). HC and CO emissions are usually very low for diesel engines owing to the fact that diesel engines operate well lean of stoichiometric. Meeting the NOx and PM standards is likely to require improved exhaust aftertreatment. Approaches under investigation include lean NOx catalysts, catalysts combined with plasma discharge systems, and particulate traps. The particle traps require regeneration by combustion of the collected soot. Exhaust temperatures must be augmented for soot combustion, either by adding a burner or by electrical heating. Adding metal additives to the fuel can lower the soot ignition temperature.

Light-duty diesel vehicles must meet emissions standards that generally are the same as those for light-duty gasoline vehicles. For some periods the diesels had less stringent NOx emission standards. Diesel vehicles are required to meet particulate matter (PM) standards whereas gasoline vehicles are generally exempt (because they have inherently low levels of PM emissions). Medium and heavy-duty vehicles The challenge for heavy-duty diesel engines is also NOx and PM control. Current engines have improved combustion chamber design, operate at lower engine speeds, use high injection pressures, use injection rate shaping, electronic control for improved timing, and use turbochargers with waste gates. Oxidation catalysts are used on many vehicles for control of the

organic carbon fraction of the particulate. Neither EGR nor particulate traps are currently used although EGR is viewed as the most promising technology for meeting future stringent NOx emission requirements. More stringent standards will require continued improvements. As with the light-duty diesel, new aftertreatment systems likely will be required. Alternative fuels are seeing greater use in heavy-duty engines than light-duty, since some vehicle fleets can be refueled at a central location. However, the use of alternative fuels in diesel engines represents a tiny fraction of the total diesel fuel use in the US.

Diesel and gasoline engines for medium and heavy-duty application must meet similar emission standards, Figure 4. United States diesel engine manufacturers recently have entered into an agreement with the USEPA to reduce NOx emissions to about 1.5 gm/kW-hr by 2004. Heavyduty gasoline and diesel exhaust emission standards in Canada are similar to the United States standards. The 1975-88 standards for diesel engines are based on the 13-mode steady state procedure, and for gasoline engines are based on the 9-mode procedure. All standards in 1989 and later are based on the USEPA transient cycle. Off-road vehicles and equipment Both direct and indirect injection diesel engines are used in off-road vehicles and equipment. Over the past 20 years, direct injection engines have been gradually replacing indirect injection engines. At the present time, indirect injection engines power only about 2% of equipment sales (USEPA, 1994). Use of turbochargers has also increased over time and currently represents about 35% of sales. Currently, about half of turbocharged engines sold are also equipped with water-jacket after-coolers. Air-to-air aftercoolers are limited to very high power output applications.

Both CARB and USEPA have implemented emission standards for engines used to power new off-road vehicles and equipment that are rated at or above 37.3 kW (USFR, 1994). The onset of Tier 1 USEPA and CARB regulations in 1996-2000 will have a relatively minor impact on diesel engine technology. Injection timing on most engines will be retarded to reduce NOx emissions. Also manufacturers may have to modify injector nozzle design, increase fuel injection pressure or turbocharger boost, add turbochargers and aftercoolers to a small fraction of sales, and add waste gates and smoke limiting devices to most turbocharged engines.

The onset of the Tier 2 CARB NOx and PM standards in 1999 and beyond are expected to lead to even more use of the techniques just mentioned above with regard to the Tier 1. Implementation of the proposed USEPA Tier 2 standards in 2001-2006 (depending on engine power rating) are expected to cause an increase in the use of electronic engine controls and highpressure injection systems. To harmonize off-road engine regulations in the United States it is likely that CARB would rescind its Tier 2 standards if USEPA follows through with the promulgation of the national Tier 2 standards. Finally, the implementation of the proposed USEPA Tier 3 standards in 2006-2008 are expected to lead to the widespread use of electronic engine controls and high pressure injection systems, as well as the use of exhaust gas recirculation.

Canada currently has no requirements or emission standards for off-highway gasoline or diesel equipment. The government is, however, redrafting existing legislation to include the authority for regulating heavy-duty off-road equipment (primarily agriculture and construction equipment),

utility engines (used in hand-held and non-hand-held equipment), and the recreational marine category (EnvCan, 1997). Mexico has yet to regulate off-road mobile sources. Military Military equipment is often powered by diesel engines or distillate-fueled turbines due to the low flammability potential of these fuels relative to gasoline. Diesel engines used in combat or tactical vehicles and equipment are designed primarily for performance, reliability and the ability to use fuel of sub-standard quality. These vehicles and equipment have been exempted from both California and Federal emission regulation. Thus, their engines incorporate little if any features designed to control emissions. Non-combat military vehicles and equipment are subject to Federal emission standards, unless they receive a national security exemption. The engines powering noncombat vehicles and engines would generally be the same as those used to power other on- and off-road diesel vehicles. Locomotives Locomotive diesel engines have long lifetimes (40 years or more), have been unregulated, and are very high emitters of NOx. Locomotives account for about 10% of mobile source NOx (EPA, 1997a). Locomotives in line haul service emit about 80 gm/kg fuel; those in switching service, about 110 gm/kg (EPA, 1997b). Initial Tier 0 emission standards apply at remanufacturing to engines built through 2001 and require a 20-25% reduction in NOx from average uncontrolled levels. Tier 2 reductions of about 55-60% apply to locomotive manufactured or remanufactured in 2005 and later.

Ships Diesel powered ships often can use either diesel or bunker fuels, in some cases switching to cleaner diesel fuel for in-port operations. NOx emissions levels are high, about 70 gm/kg of bunker fuel and 50 gm/kg of diesel fuel (BAQMD, 1997). There are no emissions regulations on ships. Bunker fuels contain high sulfur levels, making ships the only important mobile source of SO2 emissions. Ship emissions directly impact urban areas with ports. Their offshore emissions can impact upwind boundary conditions. FUELS AND FUEL USE Diesel fuel has low volatility and thus evaporative emissions are not considered a problem. Thus, there is no evaporative control system. Diesel fuels have also been reformulated. Currently, federal on-road diesel fuel can not exceed 0.05 wt. percent sulfur. In addition, diesel ignition fuel quality is controlled either directly via 40 minimum cetane number or indirectly by requiring a maximum of 35% volume aromatics. California has a separate regulation. It also requires a maximum of 0.05% weight sulfur, but, in addition, imposes a maximum of 10% volume of aromatics or certification via an engine test to demonstrate that an alternate formulation has equivalent of better emissions performance than a 10% maximum aromatics certification fuel. Nationwide, typical on-road diesel fuels now have approximately the same sulfur content as the average gasoline pool in the US. The sulfur reductions are especially beneficial for engine out particulate emissions by mitigating against excessive generation of sulfate particulate over oxidation catalysts. The European Auto/Oil study (EPEFE) has demonstrated the effects of diesel fuel reformulations for both light-duty and heavy-duty diesel engines (Hublin et al., 1996, Signer et al., 1996).

In-use emissions In-use heavy-duty vehicle emissions have been measured in tunnel studies. As of 1995, typical NO x emission factors for the in-use heavy-duty diesel fleet ranged from 30 to 50 gm per kg of fuel burned. When normalized to work output instead of fuel input, this emissions factor corresponds to the uncontrolled emission level for diesel engines of about 10 gm NOx per brake horsepower-hour. Concerns exist about the effectiveness of post-1988 exhaust emissions standards for heavy-duty engines because some engine manufacturers may be optimizing for low NO x emissions on the transient engine dynamometer certification test, and re-optimizing for maximum fuel economy under in-use driving conditions (McCracken, 1998).

Heavy-duty emissions factors are significantly higher than light-duty vehicle values of 5 to 15 gm NO x per kg of fuel. Use of control devices such as three-way catalytic converters have lowered light-duty vehicle NOx emissions, whereas only oxidation mode catalysts have been applied to heavy-duty diesel engines to control hydrocarbon emissions and odor problems. Application of three-way catalysts to heavy-duty diesel engines has not been possible to date because the excess air used during diesel fuel combustion prevents reduction of NOx in the exhaust.

ALTERNATIVE FUELS During the last two decades there has been a considerable effort in the United States to develop alternatives to the use of gasoline and conventional diesel fuel for transportation. The primary motives for this effort have been two-fold: energy security and improvement in air quality, most notably ozone. The anticipated improvement in air quality is associated with a decrease in the atmospheric reactivity, and sometimes a decrease in the mass emission rate of the organic gas emissions from vehicles using alternative fuels when compared to conventional transportation

fuels. The most common alternative transportation fuels are methanol, ethanol, compressed natural gas (CNG), liquefied petroleum gas (LPG), reformulated gasoline, and electricity.

There are a limited number of alternative fuels that can be used in compression ignition (diesel) engines (both light and heavy-duty). Vegetable oil esters can be used either neat or blended with diesel fuel in essentially unmodified engines though care is needed to assure compatibility with sealing materials and lubricants. However, the emissions benefits are at best limited and selective. Dimethyl ether (DME), which offers more attractive emission benefits, can also be used directly in compression ignition engines, but significant modification to the fuel delivery system is required to handle this gaseous fuel. Other fuels that have been used in diesel engines, namely natural gas, propane, and alcohols. All require significant engine modification, essentially converting the diesel compression ignition engine into a spark ignition engine. Overall, there are emissions advantages to some of the alternative fuels, but cost and availability considerations have limited their use.

Legislation in the United States, most notably the Alternative Motor Fuels Act of 1988 (AMFA) has accelerated the implementation and emission testing of alternative fueled vehicles (AFVs) nationwide. CARB has established a reactivity-based HC emission standard for alternative fuels to coincide with the future LEV emission standards discussed previously. Each vehicle/fuel combination will have a reactivity adjustment factor based on a quantitative scale related to the amount of ozone generated per unit mass of fuel. This requires the determination of the emission rates for all of the non-methane organic species and the associated incremental reactivity factor. The emissions-weighted average of the individual incremental reactivity factors provides an overall reactivity adjustment factor (RAF) for each fuel/vehicle combination.

JET FUEL Traditionally only that fraction of aircraft operations thought to impact urban air pollution, meaning landings, take-offs, and ground operations, has been included in emissions inventories. The altitude range considered is roughly within one kilometer of the ground, or less. The possible impact of emissions from fuel burned between 1 to 15 km is not clear but there is a possible contribution of long range NOx transport from these operations on tropospheric ozone, although it is probably of secondary importance.

About 9% of fuel used by mobile sources in the United States goes to jet aircraft. Some regions have a higher fraction. In California, 17% of transportation energy use is jet fuel (Brownstone and Lave, 1992). Most of this, nearly all of commercial operations, is Jet-A. Military operations account for about 10% of jet fuel consumption in the United States.

Over the past 30 years turbofan engines with higher efficiencies have replaced turbojet engines. Aircraft fuel efficiency [seat-km] has doubled in the past 30 years. Turboprop engines account for a small fraction of jet fuel usage, less than 2%.

Modified ground operations of aircraft can be instituted to reduce emissions during idle (reduced idle time) and taxiing (using fewer engines). An important part of airport emissions are those from ground support equipment, which should be included in off-road vehicle emissions.

Jet engines are high emitters of NOx, especially during high power operations (take-off), where emissions range from 30 to 45 gm NOx/kg fuel (Baughcum, 1996), which is 10-20 times that of a

modern automobile, Table 3. Cruise emissions are about a factor of ten lower. Total North American emissions of NOx from jet aircraft (all altitudes operation) is estimated at 0.2 MT, or less than 1% of total NOx emissions so that this source is not a major contributor. Hydrocarbon and carbon monoxide emissions are highest at low power (taxiing) operations, 0.7 to 25 gm HC/kg fuel and 10 to 40 gm CO/kg fuel. The newer engines are the lower emitting and these levels are lower than those for modern automobiles. Therefore jet aircraft are not a major source of HC or CO.

BUNKER FUELS Residual, or bunker, fuel is the very heavy, high sulfur content (3-5%) distillate left over after the refining process. It is used in ships and power plants. Since ships in international trade can carry large quantities of the fuel, relating where it is purchased to where it is used is difficult and often left out of fuel inventories.

Most of the residual fuel used in ships is used in large low or medium speed diesel engines. The remainder is used to fire boilers for steam turbine power. Ships also often burn residual fuel in boilers for heating purposes. Oxides of nitrogen levels are about 70 gm/kg fuel in diesel applications and 15 gm/kg fuel in boiler applications (BAQMD, 1997).

As with airports, ships contribute only a part of the total facility emissions. Much of the port inventory will be associated with off-road operations, including heavy-duty trucks that do not leave the facility.

OTHER ISSUES The preceding sections dealt with the nature of emissions from the mobile sector, including methods to obtain emissions factors and the relative importance of different mobile sources as can be deduced from fuel consumption and fuel based emissions data. Under “other issues” we treat the methodology for quantitative estimation of motor vehicle emissions. Specifically the progression from vehicle activity, to mobile source emissions models, to mobile source inventories. How this information is used with additional information on spatial and temporal variations as the input to ozone estimation models is reviewed. Then we examine how the validity of the mobile source emissions estimates and trends may be checked through “real world” observations using remote sensing, tunnel measurements, and atmospheric measurements. A preliminary assessment of the intersection between ozone and particulate mobile source issues provides a glimpse to the problem of dealing with mobile sources and particulates. VEHICLE ACTIVITY The common modeling approach used to produce a mobile source emission inventory is based on two processing steps: (1) determining a set of emission factors which specifies the rate at which emissions are generated (tailpipe, evaporative, or running loss emissions); and (2) determining an estimate of vehicle activity as a function of vehicle class, time of day, location, speed, and density. The emission inventory is then calculated by multiplying the results of these two steps. From this process one can obtain a time and spatially resolved emission inventory. A review of this process has been published elsewhere (Maldonado, 1991, Barth et al., 1996a, Markey, 1993). This is usually based on laboratory measurements of predetermined driving conditions. There have been several efforts to develop a more comprehensive approach to emission modeling based on vehicle operating parameters such as engine load, vehicle acceleration, vehicle speed, etc

(BDM Int., 1991, Barth and Norbeck, 1994, Barth and Norbeck, 1993). However, this approach has yet to be implemented into emission models at this time.

Vehicle activity data used for the emission inventory can come from a number of sources although it is typically produced from macroscale, regional transportation models. Traffic activity data are generated based on expected vehicle kilometers traveled (VKT), number of vehicles, number of trips, and speed distribution on a region specific basis. Hourly specific traffic densities used are usually based on measurement on the traffic network although other empirically based modeling methods are used by most traffic modelers (Warner, 1985, FHA, 1989, FHA, 1992, FHA, 1988, FHA, 1986, TRB, 1985, McGurrin and Wang, 1991). Future traffic trips and densities are based on projections of either land use, economic factors, and/or anticipated population growth (Warner, 1985, USEPA, 1997).

The travel demand forecasting process typically consists of several transportation modeling components that are used to define a network’s trip generation, trip distribution, modal choice, and trip assignment. These components are described in detail in the literature (Warner, 1985, Barth et al., 1996b). Transportation simulation modeling itself typically falls into one of two categories, microscale and macroscale. Microscale models typically model at the vehicle level and have high accuracy, but require extensive data on the system under study and require more computing power than macroscale models. Macroscale models often require less detailed data, but they sacrifice detail in order to enable the modeling of larger areas using computers with modest power. Transportation simulation models are used for analyzing various operating environments

of the road system, such as signalized intersections, arterial networks, freeway corridors, and rural highways.

Some examples of microscale performance models are TRAF-NETSIM (FHA, 1989) (for arterial networks) and FRESIM (FHA, 1992) (for the freeways). Macroscale models that are based on analytic flow models include FREFLO (FHA, 1988), TRANSYT-7F (FHA, 1986) and HCS (TRB, 1985). Many of these models were developed before the introduction of efficient and costeffective mini- and micro-computers. The models have been enhanced over the years, and many are powerful and effective. However, most are difficult to maintain and modify, contain bugs even after over a decade of development, and have rigid input/output routines structured around the punched card concept. As a result, newer traffic models are being developed in recent years that take advantage of the many developments of modeling, software engineering, and hardware platforms that have occurred over the past decade. For example, the THOREAU model makes use of object-oriented programming for greater flexibility, and is based on event-stepped simulation rather than time-stepped, resulting in greater speed performance. Other recent eventstepped transportation simulation models include the GRIDLOCK model (BDM Int., 1991), and the INTEGRATION model (Van Aerde, 1992).

Few transportation models have been combined with vehicle emission profiles, and those that do simply predict vehicle density and speed as a function of link and time to be integrated with current speed-emissions data. Although this is a step in the right direction, much better emission estimates can be achieved using a transportation model that can predict dynamic vehicle operating characteristics such as acceleration and deceleration, and combining these with modal emissions

data. Several models are being developed to integrate transportation dynamics and emissions and are described elsewhere (Barth, 1996a, Barth and Norbeck, 1994, Barth and Norbeck, 1993). EMISSIONS MODELS There are two regulatory emissions models, MOBILE, which is the Federal model, and MVEI, which is the California model. The MVEI is often referred to as EMFAC, the emission factor component of the model. Both models are structured similarly. In their simplest form they can be expressed as: Emissions [gm] = (activity data [km])(emission rates [gm/km]) The models are structured to use the emissions data generated by the certification test procedures for exhaust and evaporative emissions. Thus, the starting point for exhaust emissions is the FTPUDDS. This test has three portions: cold start, hot stabilized driving, and hot start. The model assigns model-year specific, zero-kilometer emission rates to each class of vehicles. For post1981 light-duty gasoline vehicles the rate is adjusted by technology group, i.e. how many vehicles used carburetors, throttle-body fuel injectors, or multi-port fuel injectors? This is not done for other vehicle classes. A deterioration rate is then applied to the emissions as a function of mileage. The rate increases significantly after 80,000 kilometers in the current, MOBILE5B model, although that is expected to change in the forthcoming model update. To account for the fact that some vehicles experience malfunctions or tampering that increase emissions, there are additional emitter categories, which for MOBILE5B are labeled high, very high, and super emitters. The population of vehicles in the emitter categories is increased with vehicle age. Of course, the FTP represents only one driving condition, an average speed of 19.6 mph at approximately 75 °F at a prescribed relative humidity and no road grade. A variety of algorithms have been developed to account for variations in vehicle speed and load, ambient temperature and

relative humidity, altitude, and the impact of fuels. Effects of road grade are not included in the model. Credits are also given for programs such as oxygenated fuels and inspection/maintenance.

Evaporative HC emissions modeling is also based largely on data from the certification test procedures. Thus, the models estimate and track hot soak, diurnal, running loss and resting loss emissions. The emissions are adjusted for ambient temperature, fuel vapor pressure, technology type (pre-1981, post 1981 carbureted, TBI, etc.), and I/M performance. Diurnal emissions are adjusted for the number of partial, full-day and multi-day diurnals. Running losses are also corrected for vehicle speed and trip duration. Finally, resting losses differ between open and closed bottom canisters, but not by technology type. Refueling emissions are also included in the model.

Developing adequate data on which to base the model is extremely difficult. The emission tests are expensive and time consuming. In addition, it has been recognized that the FTP test procedures have not adequately characterized in-use emissions. Recently, changes in both the exhaust and evaporative emissions tests have been made, but the resources are not available to retest the existing fleet. Acquiring random samples of the in-use vehicle population has also been a problem. As new vehicles have become cleaner, the relative importance of high emission vehicles has increased. A very large number of vehicles must be tested to properly sample the high emitters. In recognition of this problem, MOBILE5B uses IM240 emissions data to generate high emitter populations and deterioration rates for some classes of vehicles. Unfortunately, there is no comparable database for evaporative emissions. Finally, it must be recognized that accurate activity data are of equal importance to the emissions data. It has been shown that activity data

can vary significantly from county to county. Even such basic factors as the default values for the number of vehicle starts per day and the average trip length have been changed significantly due to recent studies. Other algorithms, such as the speed-correction factors, are also affected by activity. For example, an average speed of 19.6 mph is based on the distribution of speeds in urban driving captured by the FTP. Obviously, other speed distributions are possible, and perhaps likely as traffic congestion changes.

Unfortunately, evaluating the basis for the current model assumptions is difficult. The model is poorly documented and is not user friendly for research purposes. Some aspects of the models have not been updated in years, and the emission rate database for some of the vehicle categories is very poor. It is also unfortunate that there are no estimates of uncertainty in the model. Thus, it is difficult to judge how errors are propagated in the model and how accurate the overall model is. In fairness, it must be recognized that the models are frequently updated, but that resources allotted to generating the data and revising the models have been severely limited.

These models are designed to provide area wide emissions inventory information. They are not designed to provide a high degree of temporal or spatial emissions information. Nor are they designed to address issues such as the impact of traffic control measures, or to examine the emissions in local situations such as intersections. An entirely different modeling approach is needed to address these questions. Currently several groups are working on the development of modal emissions models. The approach taken in these models is to develop emission rates on the basis of vehicle operating modes, then to aggregate the modes to fit the modeling domain. This requires that emissions data be collected on a sec-by-sec basis along with a variety of vehicle

operating parameters. Efforts are still underway to determine how many modes are required to properly characterize emissions and to build the models.

A third approach is fuel-based modeling. This approach, which is examined in some detail in this paper, uses emissions per volume of fuel used rather than emissions per vehicle kilometer traveled. This has several advantages. First, fuel consumption data are readily available on an area wide basis. Thus, if the emission rates can be determined, inventories can be easily calculated. Second, it has been shown that pollutant emission rates are much more constant on a fuel basis than on a distance basis. For example, emission rates per kilometer traveled determined in underwater vehicle tunnels varied by a factor of two depending on whether traffic was going downhill or uphill (Pierson et al., 1996b). On-road measurements show similar results (CiceroFernandez et al., 1997). On a fuel basis, however, emission rates were approximately constant. Third, remote-sensing data are directly applicable to fuel base modeling, as would be roadside studies that measure pollutant ratios to CO2. The drawbacks to this approach are that start, diurnal, hot soak, and resting loss emissions are not only related to the quantity of fuel used but also how the depend on vehicle is operated. Furthermore, fuel sales data can not be used to provide the spatial or temporal data needed for some studies.

There are at least four emission models that are in operation to estimate on-highway vehicle emission inventories in North America. These models are listed below: MOBILE5B (United States, other than California) MOBILE5C (Canada) EMFAC7 (California)

MOBILE5J (Juarez, Mexico) In addition, other approaches based on modal analysis (depending upon the operating condition of the vehicle) and fuel consumption have been proposed. MOBILE5B MOBILE5B is USEPA's current model for estimating emission rates for light- and heavy-duty vehicles. It produces HC, CO, and NOx emissions in g/mi for seven vehicle types, and has the ability to adjust these emission rates for varying speeds, temperatures, and control programs, such as Inspection/Maintenance programs and reformulated gasoline. USEPA is currently revising MOBILE5B and plans to release MOBILE6 sometime in 1998. MOBILE5C is Canada's version of the MOBILE model for Canadian vehicles. Major differences in MOBILE5C and MOBILE5B are the reflection in MOBILE5C of Canadian emission standards that are substantially different than the United States standards, especially for pre-1988 vehicles, and Canadian vehicle kilometers traveled versus age and registration versus age distributions. EMFAC7 is used in California by the various Air Quality Districts, and by the Air Resources Board. This model is similar in many respects to MOBILE5B, but also differs substantially in calculation methodologies. For example, EMFAC estimates cold start emissions in grams per trip, while MOBILE5B estimates cold start emissions in g/km. There are also substantial differences in emission rates of vehicles that are almost identical. This is something that the USEPA is addressing in the next generation of their model, MOBILE6. Finally, the MOBILE version for Mexico-Juarez is a model that was developed by Radian Corporation under contract to USEPA based on the in-use vehicle emission factors measured in Juarez by the USEPA and by the State of Texas. Mexican officials, however, are in the process of creating a number of MOBILE5 derivatives for use in various parts of Mexico (Espinosa et al., 1996).

Emission rates in g/km for various vehicle classes from MOBILE5B are shown in Table 4 below. To obtain these emission rates, MOBILE5B was run at 24 oC, without an inspection maintenance program, anti-tampering program, or reformulated gasoline. Operating modes were standard defaults, gasoline fuel volatility was assumed to be 63 kPa Reid vapor pressure (RVP) and minimum and maximum temperatures were 16 oC and 29 oC. Hydrocarbons are shown as total organic gas or, TOG, which includes methane, ethane, and aldehydes. Hydrocarbon emissions include evaporative, as well as exhaust emissions. Fleet average emission rates are shown for 1997 and 2010. The “all vehicles” emission rates are not a sum of the other emission rates, but are the fleet emission rates weighted by vehicle kilometers traveled fraction of each vehicle class. The results show 1997 TOG and NOx emission rates of 1.83 and 1.58 g/km, respectively. The TOG and NOx emission rates are projected to drop by 24% and 21% to 1.39 and 1.24 g/km, respectively, by 2010. These fractions may change significantly with MOBILE6.

Using the vehicle kilometers traveled fractions for each vehicle class, the fraction of TOG and NO x emissions that was due to gasoline vehicles, diesel vehicles, and motorcycles was estimated. These results are shown in Table 5 for both 1997 and 2010. In 1997, gasoline vehicles (excluding motorcycles) account for 93% of the TOG, and 65% of the NOx. Diesel vehicles account for only 6% of the TOG, but 35% of the NOx. The TOG fractions are approximately the same for 2010, but the fraction of NOx due to gasoline vehicles increases to about 72% in 2010.

TOG emissions include both exhaust and evaporative emissions. Evaporative emissions components are running losses, hot soak emissions, diurnal emissions, resting losses, and

refueling losses. Evaporative emissions are sensitive to ambient temperature. The higher the temperature, and the greater the spread between daily minimum and maximum temperatures, the greater evaporative emissions become.

A comparison of exhaust and evaporative TOG emissions at three different temperatures for light-duty gasoline vehicles is shown in Table 6. Exhaust emissions were estimated at the average temperature listed, and the model using the low and high estimates listed estimated evaporative emissions. The 24 oC and 16-29 oC temperatures are the temperatures used in testing vehicles over the Federal test Procedure (FTP). The 35 oC and 22-36 oC temperatures are used to represent average ozone violation days in non-California areas of the United States, and the 41 oC and 21-41 oC temperatures may represent an ozone violation day in the interior parts of Southern California.

At lower summertime temperatures, evaporative emissions are 23-29% of total TOG emissions. However, at higher temperatures, exhaust emissions increase slightly, while evaporative emissions can increase by a factor of 3-4.

The on-highway gasoline fleet includes two basic groups of vehicles--normal emitters, whose emissions are relatively low and non-normal emitters, whose emission rates are relatively high. The USEPA and California divide high emitters into different classes by emission levels, but basically, they can all be thought of as non-normal emitters. To illustrate the contribution of nonnormal emitters to fleet emissions for passenger cars, normal emitter emission rates were obtained from the USEPA and input into the MOBILE5B for comparison with the overall emission rate

(which combines normal and non-normal emitters). The fraction of emissions due to non-normal emitters is shown in Table 7. Over one-half of the HC and CO emissions for the 1997 fleet of passenger cars is attributable to non-normal emitters. It is somewhat less for NOx. In 2010, the percent of emissions due to non-normal emitters rises to almost 70% for HC and CO, and 34% for NO x. Comparison of models MOBILE5B emission rates were compared to emission rates from the other three models used in North America: MOBILE5C (Canada), EMFAC7G (California), and MOBILE5-Juarez (Mexico MOBILE5 for Juarez) (RITNRCC, 1996). In general, emission rates by model year were higher in Juarez than the United States due to differences in emission standards, and also due to the high rate of catalyst poisoning (leaded fuel was phased-out in 1994). MOBILE5C was run with Ontario inputs, and EMFAC7G was run to estimate a California statewide planning inventory. The comparison for 1997 and 2010 is shown in Table 8.

MOBILE5B and MOBILE5C show very similar emission rates for both 1997 and 2010. MOBILE5C has higher emission rates for pre-1987 vehicles, but the Ontario fleet mix is generally newer than the MOBILE5B fleet mix, thus the emission rates are similar. MOBILE5C NOx is lower than MOBILE5B, primarily due to the newer fleet mix in MOBILE5C. EMFAC7G has much lower emission rates for California, both for 1997 and 2010. For 1997, this is due primarily the fact that EMFAC7G has much lower deterioration rates for 1983 and later light-duty gasoline vehicles and light-duty gasoline trucks than MOBILE5B, and also to the lower volatility of California reformulated gasoline. For 2010, however, the lower California emission rates are due to the California Low Emissions Vehicle (LEV) program, and onboard diagnostics (OBD) controls

required on light- and heavy-duty vehicles. These OBD controls are also required on Federally certified cars and trucks, but MOBILE5B does not currently include the effects of onboard diagnostics. The MOBILE6 model, which EPA is developing now, most likely will include lower emissions for vehicles equipped with onboard diagnostic control systems. The MOBILE5B emission rates for Juarez for hydrocarbons are significantly higher than the United States and Canada emission rates. However, NOx appears to be somewhat lower. The reason for the lower NO x is a much lower diesel fraction in Juarez than in the United States. For example, the fraction of total vehicle kilometers traveled for heavy-duty diesel vehicles (i.e., those over 3860 kg gross vehicle weight) in Juarez is about 2%, where in the United States it is typically between 6% and 8%, depending on the calendar year. Comments on MOBILE6 The USEPA is working on the next version of the MOBILE model. Changes discussed to date include: addition of credits for remote sensing and other in-use control programs, adding benefits of the Federal OBD-II requirements, using 24-hour diurnal evaporative emissions data to replace the current data based on accelerated one-hour tests, inclusion of a new emissions category to handle gross liquid leakers, updated heavy-duty emission factors, revised effects of fuel oxygen on exhaust CO, inclusion of the effect of sulfur on exhaust emissions, addition of the effect of air conditioning on emissions, inclusion of facility specific off-cycle emissions, and revised deterioration rates for running exhaust emissions. Some of the current deficiencies, such as the failure to treat grades, will probably remain.

EMISSION INVENTORIES National mobile source emission estimates were developed as part of this review, using fuel sales data for each country, and representative emission factors for NOx and NMOC measured in tunnel studies (see Table 9). Typical values of 10 and 40 g NOx per kg of fuel burned were used for gasoline and diesel fuel, respectively. Fuel energy content was converted to mass units using lower heating values of 44 and 43 MJ/kg for gasoline and diesel fuel, respectively (Heywood, 1998). For Canada, total mobile source NOx emissions were estimated to be ~1 Tg, with roughly equal contributions from on-road diesel, off-road diesel, and gasoline engines. The off-road contribution in Canada is somewhat higher when NOx emissions due to combustion of jet fuel by aircraft and residual fuel oil in marine engines are included. For the US, total mobile source NOx emissions were ~10 Tg, again roughly equally divided among gasoline engine sources, on-road diesel, and heavy-duty off-road sources (diesel, jet, and marine engines). In both the US and Canada, gasoline engines are responsible for >90% of mobile source NMOC emissions. This conclusion is supported by the NMOC emission indices shown in Table 9, and strengthened by noting that additional NMOC emissions not reflected in Table 9 occur due to gasoline evaporation. Furthermore, the NMOC emission factors shown in Table 9 for HD vehicles are upper bound values when applied to diesel engines, because these measurements included contributions from HD gasoline engines. Suitable emission factors for Mexican mobile sources are not yet available.

Emission inventories of VOC and NOx were examined in the United States to determine the contributions of on-highway and off-highway to the total mobile source inventory for VOC and NO x. These inventories are shown in Table 10 below. The source of the United States inventories

is USEPA's 1995 Emission Trends Report (USEPA, 1996). The United States inventories are for calendar year 1995, and the California inventories are for 1997. The United States inventories are estimated under annual average, rather than ozone season conditions.

For the United States, on-road sources in 1995 are estimated to contribute about 73% and 79% of the total mobile source VOC and NOx, respectively, on an annual average basis. As a whole, mobile sources (on-road + off-road) contribute 38% of the VOC and 42% of the NOx in the United States (again, annual average basis). However, important geographical and seasonal differences exist. In southern California during the summertime, mobile sources are estimated to contribute 53% of the VOC and 86% of the NOx. We believe that these “official” inventories underestimate mobile source contributions, particularly light-duty vehicle VOC and CO and heavy-duty NOx. TYING EMISSIONS TO OZONE Spatial and temporal distribution of emissions Mobile source emissions must be inventoried on the spatial and temporal scales over which tropospheric ozone pollution problems are observed. This requires that inventories be resolved within urban and regional spatial scales, by hour of the day, for summer days when conditions are conducive to ozone formation in Canada and the United States, and year-round in Mexico City.

Annual fuel use data for mobile sources are readily available at the national level in Canada, Mexico, and the United States (see Figure 1). Usually fuel use can be resolved to the state or provincial level and by month as well. Market surveys in which gasoline sales are measured directly at service stations are conducted routinely in the United States and Canada; such data can

be used to apportion gasoline sales to subregions within a state or province. It is more difficult to apportion spatially the diesel fuel consumption, because the point of sale can be widely separated from most of a long-haul diesel truck trip (large diesel trucks can travel up to 1600 km between refueling.

For on-road vehicles, the influence on emissions of monthly variations in fuel sales is small compared to the more important seasonal factors such as changes in temperature and variations in gasoline properties such as vapor pressure. Fuel use by off-road mobile sources is not well known, because off-road mobile sources are rarely subject to registration or licensing requirements. Seasonal variations are expected in activity levels of off-road mobile sources, for example in the agriculture sector.

Determining fuel use and vehicle emissions on fine spatial scales is a challenging task. It is common practice to use travel demand models to predict motor vehicle activity within the urban scale (Harvey and Deakin, 1993, TRB, 1995). Travel demand models make use of socioeconomic data such as population, employment, automobile ownership, and household income, combined with information about travel times between points, available modes of transportation, and a description of the roadway network. Such models predict spatially and temporally resolved vehicle activity, which can be combined with emission factor model predictions (discussed below) to develop the overall emission inventory. Travel demand models can be used to predict future traffic volumes based on forecasts of socioeconomic conditions and land use.

The goal of travel demand modeling has been to estimate total traffic volumes, which are dominated by light-duty passenger vehicles. Travel demand models do not describe heavy-duty truck travel explicitly; often truck travel is estimated as a constant fraction of total traffic volumes. However, diesel truck activity does not follow the same pattern as passenger vehicle travel. On weekdays within urban areas, diesel truck travel peaks around midday, and falls off before the afternoon commuter peak period (Schlappi et al., 1993). The effect on ozone of greater heavy-duty diesel NOx emissions during the middle of the day has not been assessed.

Day-of-week differences in ozone concentrations have been observed: in some urban areas, average ozone concentrations are higher on weekends (Altshuler et al., 1995). Differences in mobile source emissions are likely to contribute to this phenomenon. Diesel truck activity and emissions within urban areas decrease dramatically on weekends (Dreher and Harley, 1998). In contrast, the total amount of passenger vehicle travel is similar on weekends and weekdays. Related observations showed a small hourly variation in long-haul diesel operation but a large hourly variation in passenger car operation. Travel demand models have a strong employmentdriven component to their predictions, and so require modifications before they can be used to assess weekend travel.

An alternate approach to defining vehicle activity involves direct observations of traffic volumes by automated counters. Many highways are now heavily instrumented, and it is possible to extract hourly traffic counts and speed distributions on each highway segment. Automated traffic counters also are being installed on urban arterial roadways. Weigh-in-motion sensors provide information on heavy-duty truck travel. Where possible, direct measurement of vehicle activity,

especially for heavy-duty trucks and for all vehicles on weekends, appears preferable to relying on travel demand model predictions. VOC speciation and reactivity Roadway tunnel measurements have been used to define the detailed chemical composition of mobile source VOC emissions (Lonneman et al., 1986, Zielinska and Fung, 1994, Kirchstetter et al., 1996, Sagebiel et al., 1996, Rogak et al., 1997a). Numerous dynamometer studies of exhaust VOC composition also have been reported (Sigsby et al., 1987, Hoekman, 1992, Hochhauser et al., 1992). Analyses of both dynamometer and tunnel data have concluded that unburned gasoline constitutes 50% or more of total VOC exhaust emissions (Leppard et al., 1992, McLaren et al., 1996a). Weight fractions of combustion-derived compounds such as acetylene and ethene in exhaust have been declining over time, while the ratio of ethene to acetylene has been increasing (comparisons of engine-out and tailpipe-out emissions confirm that catalytic converters are especially effective in removing acetylene). Relative to United States studies, Canadian vehicle VOC emissions measured in the Cassiar Connector (Gertler et al., 1996, Rogak et al., 1997a) show much higher propane, and lower aromatic fractions. Likely factors contributing to these differences include the presence of more LPG-powered vehicles, and less use of aromatics in gasoline because methylcyclopentadienyl manganese tricarbonyl (MMT) is allowed as an octaneimprover in Canada.

For many vehicles, differences in the composition of VOC emissions are found during the first few minutes of vehicle operation when the engine and catalytic converter are cold. However, cold start emissions are similar in composition to the hot stabilized VOC emissions of high-emitting

vehicles that are operating fuel-rich, with little or no catalytic converter activity (Kirchstetter et al., 1996).

The speciation of evaporative VOC emissions is commonly represented using the composition of liquid gasoline and its headspace vapors. Relative to liquid gasoline, headspace vapors are enriched in C4–C6 compounds. The composition of various vehicle-related evaporative emission sources also has been measured directly (Burns et al., 1992).

Zielinska et al. (1994) note that for HD diesel exhaust, a significant fraction of VOC masses is emitted in the form of C10 and larger molecules. Therefore, stainless steel canister and Tedlar bag sampling are inadequate for characterizing diesel exhaust VOC emissions. Full heavy-duty diesel exhaust speciation has been derived from tunnel measurements (Sagebiel et al., 1996, Zielinska et al., 1996).

Carter (1994) has described reactivity scales that can be used to compare the reactivity of different VOC with respect to ozone formation. The maximum incremental reactivity (MIR) scale has been widely used for this purpose. Specific or normalized reactivity is computed as a weighted average using weight fractions and MIR values for all individual VOC present in an emissions sample; the result is reported in units of g O 3 per g NMOC. For any given speciation profile, the absolute assessment of reactivity is subject to numerous assumptions and modeling uncertainties. Conclusions about reactivity are more robust when they are made in a relative sense (e.g., comparing the reactivity of exhaust or evaporative emissions using two different gasoline formulations).

Specific reactivity of light-duty vehicle exhaust emissions calculated using the MIR scale ranges from 2-3 g O3 per g NMOC emitted in dynamometer studies (Hoekman, 1992, Hochhauser et al., 1992). On-road determinations of exhaust reactivity in tunnels suggest somewhat higher values of ~4 g O3 per g NMOC (Kirchstetter et al., 1996, Sagebiel et al., 1996). Calculations for the Canadian light-duty vehicle fleet using speciation profiles presented by Rogak et al. (1997a) suggest lower specific reactivity of ~3 g O3 per g NMOC, possibly due to lower aromatic hydrocarbon levels in Canadian gasoline.

Changes in gasoline formulation affect the speciation and in some cases, the reactivity, of VOC emissions. In general, gasoline reformulation has small effects on the reactivity of exhaust emissions, but can have larger effects on the reactivity of evaporative emissions (Hoekman, 1992, Hochhauser et al., 1992, Burns et al., 1992). Removing heavy aromatics can lower exhaust reactivity. Reactivity of liquid gasoline is reduced by adding MTBE and by removing olefins and aromatics. The reactivity of headspace vapors is reduced by adding MTBE and by removing C4C6 olefins. Kirchstetter et al. (1997) reported that California's Phase 2 reformulated gasoline reduced the specific reactivity of both liquid gasoline and gasoline headspace vapors by ~20%. The reactivity of exhaust VOC emissions decreased by ~5%. Alternate fuels such as natural gas, LPG, and methanol, can reduce the specific reactivity of exhaust VOC emissions by factors of 2 or more. Assessment of the ozone-forming potential of VOC emissions must consider both the mass emission rate and the specific reactivity; changes in total VOC mass emission rates due to use of reformulated fuels are considered elsewhere in this review. The reactivity benefits of

reformulated fuels are of minor importance in cases where tropospheric ozone is limited by NOx emissions. REMOTE SENSING Much has been learned about in-use motor vehicle emissions through use of spectroscopic remote sensing techniques that measure emissions from individual vehicles as they drive by roadside sensors (Stephens and Cadle, 1991, Bishop et al., 1989, Guenther et al., 1995, Zhang et al., 1996, Zhang et al., 1993). Vehicle emissions have been measured using remote sensors in numerous United States cities, as well as in Canada, Mexico (Beaton et al., 1992, Bishop et al., 1997), and other locations around the world (Zhang et al., 1995). Among the advantages of remote sensing techniques are: the ability to measure emissions from large numbers of vehicles, the ability to monitor vehicles in their normal in-use operating condition, and reduced sampling bias because drivers are not notified of the testing or given the option of not participating. The sample is usually restricted to those vehicles with tailpipes near roadway level, although limited measurements of diesel truck emissions have been made with a remote sensor elevated ~5 m above roadway height (Bishop et al., 1996). While some long-pathlength spectroscopic measurements of aircraft emissions have been made across airport runways, in general there has been little use of remote sensing techniques to monitor off-road mobile source emissions.

Since the light beam passes through only a small portion of the vehicle exhaust plume, it is not possible to obtain an absolute emission rate with this technique. Instead, a large number of simultaneous concentration measurements are made of both CO2 and the species of interest within a one-second time interval. The measurements are used to calculate the ratio between the pollutant and CO2. This approach has two advantages. First, if the ratio is not constant during

the measurement period, the data are suspect and can be rejected. Second, if both CO and CO2 are monitored, then almost all of the carbon that was present in the combusted gasoline has been measured. By using an average gasoline C:H:O ratio and density, the emission rate of the pollutant can be calculated in either units of grams-per-gallon of gasoline consumed or percent concentration in the raw exhaust. Thus, the data can be directly used in fuel-based inventory modeling.

Infrared remote sensing capabilities are best developed for CO. Qualitatively, remote sensing techniques are able to identify vehicles with high exhaust HC emissions; evaporative emissions are not measured. Remote sensing measurements of HC emissions are difficult to interpret quantitatively because hundreds of different compounds are present in vehicle exhaust and each compound has a different infrared absorption spectrum (Stephens et al., 1996). The standard HC channel on infrared remote sensors is at 3.4 µm, where alkanes absorb strongly. Other classes of HC such as aromatics absorb weakly at this wavelength. Various approaches have been developed to measure NOx emissions: ultraviolet absorption (Zhang et al., 1996, Popp et al., 1997), broadband infrared absorption with correction for interference by water vapor (Jack, 1997), and tunable infrared diode laser absorption spectroscopy (Koplow et al., 1997).

A number of validation studies have been performed in which remote sensor data have been compared to calibration gas mixtures, emissions from passing vehicles with on-board exhaust concentration measurement capability, and the emissions from vehicles which have been independently measured using both inspection/maintenance and certification measurement procedures. These studies have confirmed that remote sensors are capable of providing excellent

measurements of on-road CO and HC emissions. The accuracy of the NOx emissions data is not as well defined at this time. There remains, however, controversy over how best to use remote sensing data. Much of the debate is over how well a one second remote-sensing measurement characterizes the emissions from an individual vehicle.

Vehicle emissions can be highly variable on both a short and a long-term basis. Short term variability is caused by changes in operating mode (i.e. throttle dither), enrichment due to highload events, rapid changes in catalyst efficiency during light-off, and erratic emission control due to emission control component failures or tampering. Long-term variability is caused by vehicle deterioration, changes in fuel, and failure mode operations. A failure mode that results in rich operation can easily result in a 100-fold increase in a vehicle’s HC or CO emission rate. Large differences also exist between individual vehicles of different ages due to improvements in the emission control technology. Understanding this emission variability is the key to properly using and interpreting remote sensing. First, the high HC and CO emission rates associated with cold start operation and high-load events can be avoided by careful selection of the remote sensing location. Second, one must recognize that while emissions during hot-stabilized operation from properly functioning vehicles vary greatly sec-by-sec, their emission rates rarely approach those that are typical for a high emission vehicle. Third, some high emission vehicles can experience very wide changes in emission rate, such that they appear to be low emission vehicles one second and high emission vehicles shortly thereafter. Overall, this variability makes it difficult to use remote sensing as a highly accurate tool for identifying high emitting vehicles. Two approaches are under investigation to minimize errors in separating high and low emitters. One is to identify

high emitters based on multiple remote sensing measurements. The second is to identify only low emitters, also based on multiple remote sensing measurements.

For inventory purposes, the issue of emissions variability is moot. A single remote sensor can measure the emissions from thousands of vehicles per day. By averaging the emission rates from large numbers of vehicles, the sec-to-sec variability issue is removed, although one must still be careful of biasing the data by selecting a site with atypical vehicle operation or an atypical fleet. Simultaneous recording of license plate information combined with automated license plate readers, makes it practical to identify most vehicles by model year or even make and engine type through state registration data. A large number of remote sensing studies has now been conducted in the United States and elsewhere. In the United States, these studies show that a small percentage of the in-use vehicle fleet contributes the majority of the exhaust emissions. Typically, 10% of the fleet is responsible for 50% of the emissions. Nine to 11 year old vehicles typically contribute the most emissions to the inventory. The impact of older vehicles is decreased due to scrappage and decreased VKT.

Remote sensing appears to be well suited to long term monitoring programs. By establishing remote sensing sites that are used periodically over a period of years, one could directly monitor the emissions changes in the in-use fleet. Unfortunately, such sites have not been established. The only long-term site to date is one that the University of Denver has used in several programs. Analysis of data at that site suggests that vehicles manufactured since 1985 are experiencing a reduced rate of emissions malfunctions, and that the improvement is further enhanced for 1991 and later model year vehicles. The Coordinating Research Council is currently establishing a series

of remote sensing sites for a five-year study. There is a need, however, to extend such monitoring program indefinitely. TUNNEL STUDIES Roadway tunnels are being used to measure on-road vehicle emissions. Measured emissions in tunnels have been compared to emission factor model predictions for tunnels in Van Nuys, CA (Pierson et al., 1990), Tuscarora, PA and Baltimore, MD (Pierson et al., 1996b, Robinson et al., 1996), and Vancouver, Canada (Gertler et al., 1997b, Rogak et al., 1997a). Due to uncertainties in tunnel air flow measurements and in the vertical profile of pollutant concentrations at tunnel inlets (Rogak et al., 1997b), it is common practice to consider ratios of pollutant emission rates (e.g., CO/NOx) which are less sensitive to measurement and modeling uncertainties.

It is especially useful to normalize measured pollutant concentrations to total exhaust carbon (i.e., the sum of CO 2, CO, and VOC). Table 9 summarizes on-road measurements of emission rates for CO, NMOC, and NOx, expressed per unit mass of fuel burned based on the normalization to total carbon. Light-duty (LD) vehicles are almost entirely gasoline-powered, whereas heavy-duty (HD) vehicles are >70% diesel-powered. Based on the emission indices shown in Table 9, and knowing that gasoline sales are several times greater than diesel fuel sales by mass, it is clear that HD vehicles are a minor source of CO and NMOC, and a major source of NOx emissions. LD vehicle emissions in Canada, as measured in the Cassiar Connector in Vancouver, are within the range of values seen in United States tunnels, with the exception of NOx, which is higher. Rogak et al. (1997a) report that differences in new-vehicle emission standards are unlikely to explain the higher NOx emissions observed for Canadian LD vehicles. Further study of this issue is needed.

As a method for characterizing on-road vehicle emissions, tunnel studies have both advantages and disadvantages when compared to laboratory testing. The main advantage is that emissions are measured from a large random sample of in-use vehicles operating under real-world conditions. Such sampling is needed to ensure that gross-polluters are adequately represented in the emissions determination. Parameters such as distribution of model years, vehicle types, and average speed can be measured directly in tunnel studies; thereby reducing vehicle activity-related uncertainties in the comparison of tunnel results with emission factor model predictions.

However, tunnels do not capture the full range of vehicle emissions. For example, there is little influence of cold start or evaporative emissions on most tunnel measurements (of the evaporative sources, only running losses are well represented). High-speed and heavy acceleration driving conditions may not be encountered inside some tunnels. Other relevant variables such as ambient air temperature, use of vehicle air conditioning, and fuel composition are not readily controlled in tunnel experiments.

Whereas emissions expressed per unit distance traveled vary significantly with driving conditions, emissions normalized to fuel consumption vary much less, as can be seen from Table 9 by comparing Fort McHenry results for uphill versus downhill driving (Pierson et al., 1996b). This suggests that normalizing emission factors to fuel consumption will provide a more robust basis for estimating mobile source emission inventories, compared to current models that are based on distance traveled.

For LD vehicles, both tailpipe and some evaporative emissions are found in tunnels. Evaporative emissions are estimated to account for 10-15% of total NMOC emissions in both United States and Canadian tunnel studies (McLaren et al., 1996a, Gertler et al., 1996). Evaporative emissions are negligible for diesel vehicles, because of the low volatility of diesel fuel. ATMOSPHERIC MEASUREMENTS AND SOURCE RECONCILIATION Concentrations of CO and NOx are measured routinely at North American air monitoring sites. Ambient VOC concentrations also are measured, though the spatial and temporal coverage of the monitoring is much less complete than for CO and NOx. Continuous monitoring of total nonmethane organic compound (NMOC) concentrations should be an integral part of air monitoring networks. This information can be augmented by occasional measurements of full NMOC speciation. Long-term records of ambient concentration data are useful in assessing emission trends.

Ambient concentration data also provide a useful independent check on mobile source emission inventories. A widely cited top-down (starting with ambient concentrations) analysis of the mobile source emission inventory has been presented by Fujita et al. (1992) for southern California. Fujita et al. assumed NOx emissions were estimated correctly, and then determined from measured ambient CO/NOx and NMOC/NOx ratios that CO and NMOC emissions in 1987 were understated in mobile source inventories by factors of 1.5 and 2.0-2.5, respectively. A critical review by Yarwood et al. (1994) notes that the acceptability and accuracy of the NO x data is the cornerstone of the study by Fujita et al.; potential problems with the NOx ambient and emissions data are the weak link in the conclusions about the CO and NMOC inventories. Harley et al. (1997) used NOx/CO and NMOC/CO ambient ratios, combined with a fuel-based estimate

of CO emissions (Kirchstetter et al., 1996, Singer and Harley, 1996), to estimate absolute NMOC and NOx emissions in southern California. While MVEI 7G model predictions for NOx were consistent with top-down estimates made by Harley et al., NMOC emissions remained underestimated by a factor of 2.4. Clearly, large uncertainties remain in NMOC emissions estimates for historical ozone episodes; top-down assessments of mobile source emission inventories are needed urgently for present-day conditions.

Receptor modeling techniques permit the back-calculation of source contributions to total NMOC concentrations. A review of receptor modeling studies has been prepared by Pierson et al. (1990) and Yarwood et al. (1994). Within and across many urban areas in the United States and Canada, the composition of NMOC in ambient air samples is similar (Fujita et al., 1992, Fujita et al., 1995, Kenski et al., 1995, McLaren et al., 1996b, Jiang et al., 1997). Receptor modeling calculations have found mobile source-related emissions to account for ~70% of ambient NMOC in many urban areas (Harley et al., 1992, Fujita et al., 1994, Fujita et al., 1995, Fujita et al., 1997, Lin and Milford, 1994). Mobile source contributions to NMOC are often higher than corresponding inventory estimates suggest. Notable special cases are: Mexico City where LPG emissions have been identified as a significant source (Blake and Rowland, 1995); Bakersfield, CA and environs where oil field emissions were important (Fujita et al., 1995); and the Detroit, MI area where contributions from refineries were higher than expected (Scheff et al., 1996).

Receptor modeling studies often seek to separate the exhaust and evaporative contributions to total mobile source NMOC emissions. While tailpipe exhaust is the dominant source, additional contributions from liquid gasoline and headspace vapor sources also have been identified.

Separation of whole gasoline evaporation from tailpipe sources is difficult because tailpipe emissions include unburned fuel that escapes combustion. Source contribution estimates for whole gasoline are sensitive to the source composition profile used for exhaust emissions, especially the weight fractions of combustion-derived species such as acetylene and ethene (Fujita et al., 1994). Henry et al. (1994) exploited the correlation structure in ambient data to deduce not only source contributions, but also the source composition profiles themselves. They acknowledged the presence of unburned gasoline in tailpipe emissions, but concluded that "there are additional source(s) of whole gasoline unconnected with vehicles in motion." In receptor modeling for Boston, MA and Los Angeles, CA in 1995, the contribution of whole gasoline was lower, and that of headspace vapors higher, than reported in previous studies. Fujita et al. (1997) suggest that substitution of fuel injectors for carburetors may help to explain this trend. INTERSECTION OF OZONE AND PM MOBILE SOURCE ISSUES Mobile sources contribute to the atmospheric burden of particulate matter (PM) through three mechanisms, primary PM emissions, secondary PM formation, and fugitive emissions. Primary particles are those directly emitted by vehicles. They are emitted in the vehicle exhaust, and also are emitted via brake and tire wear. The PART5 model is the United States USEPA’s official model for projecting PM emissions from on-road motor vehicles. This model uses the same vehicle classes and activity data as the MOBILE5 model, and estimates emissions from both light-duty and heavy-duty vehicles (Rykowski et al., 1996). However, the algorithms for PM are much simpler than those used for other emissions. For example, there is no separation of cold start from other operating modes and no deterioration of emission factors. This is largely due to a lack of data. Several studies have recently, or are currently, examining light-duty gasoline and heavy-duty diesel in-use PM exhaust emissions. The results will be used to improve current

model estimates. Both tire and brake wear PM emissions factors are out of date and need to be addressed. The greatest changes have probably occurred with brake wear PM emissions, since brake pad materials have been changed since the earlier studies. Primary PM emissions are not known to have an impact on ozone, although they contribute to the atmospheric pool of fine particles that are available for heterogeneous chemistry.

Secondary PM formation comes via emissions of HC, NOx, and SO2 which then react in the atmosphere to give organic, nitrate and sulfate PM, respectively. Roughly 30-50% of all PM is estimated to be due to secondary inorganic PM and ca. 5% due to organic secondary PM. The formation routes of secondary organic PMs are not well quantified. However, it is believed that C7 + HCs react in the atmosphere (if it contains ozone) to form less volatile organic species that condense into the particle phase, thereby contributing to the carbonaceous particle burden (Odum et al., 1997, Grosjean, 1992, Wang et al., 1995, Turpin and Huntzicker, 1995, Schauer et al., 1996). The rate of secondary organic PM formation from aromatic compounds is believed to be roughly twice as fast as those from alkanes and alkenes. NOx oxidizes to nitric acid, which reacts with ammonia to form ammonium nitrate. The amount of ammonia in the atmosphere can be the controlling factor in the amount of particulate nitrate formed. SO2 is oxidized to sulfuric acid, which exists primarily in the particle phase. Mobile sources are minor contributors to the SO2 in most areas but are often major contributors to NOx and C7+ HCs. Since ozone is formed via atmospheric reactions involving HC and NOx, improving our understanding of the mobile source HC and NO x inventory will increase our knowledge of the ozone and PM mobile source issues and how they interact.

Fugitive emissions are PM that is resuspended by the turbulence caused by vehicle traffic. The amount of dust generated is a function of the amount of fine particles (silt) on or near the roadway. This is a major source of PM-10 (particles with a diameter less than or equal to 10 µm), but will be less important for the new PM-2.5 (particles with a diameter less than or equal to 2.5 µm) standard since reentrained dust is mostly large particles. As with primary PM emissions, any overlap with the ozone issue comes in the area of heterogeneous atmospheric processes.

PROJECTING THE FUTURE The uncertainties in assessing current mobile source emissions are compounded when estimating future year emissions. Because of the slow turnover of the motor vehicle fleet, especially heavyduty vehicles, the impact of new technology is felt slowly. Fuel changes, if they occur and apply to current vehicles, have a more immediate effect. Advances in technology and fuels tend to be offset by growth in population, fuel consumption, numbers of vehicles, and vehicle kilometers driven. Attempts to constrain or reduce private vehicle use through transportation control measures (TCMs) have met resistance and produced minimal results.

Projecting future in-use fleet average emissions is difficult under the best of situations. Doing so with models that have difficulty projecting current emissions adds to the uncertainty. As discussed, current emissions are heavily influenced by the emissions of high emitters that make up a relatively small fraction of the fleet. Since the emissions from properly functioning new vehicles are continuing to decrease, the relative importance of malfunctioning vehicles is likely to grow. Off setting this trend are programs designed to identify and repair high emitters. Thus, future emissions estimates must account for changes in vehicle durability, changes in emissions

related tampering, effectiveness of inspection/maintenance programs, the effectiveness of the OBDII system, and human behavior in terms of the willingness of people to maintain their vehicles in a timely manner. In addition, estimates have to be made of the in-use effectiveness of new emissions regulations, and socio-economic issues such as the rate of fleet turnover. Factors that need to be considered for the United States, and in some cases for Canada and Mexico, are: •

Federal Tier I exhaust emissions standards were in effect for all new vehicles starting in 1996. Their in-use performance needs to be monitored.

•

Tier I light-duty vehicle (passenger cars) exhaust standards extend to 10 years or 160,000 kilometers as opposed to the 5 years or 80,000 kilometers for earlier standards. This may impact in-use durability.

•

The California TLEV, LEV, ULEV, and ZEV exhaust standards will phase in between 1993 and 2003. The performance of these vehicles needs to be monitored.

•

NLEV standards will phase in starting in the northeastern states in 1999 and will go national in 2001.

•

Federal Tier II exhaust standards may be implemented in 2004.

•

The requirement to meet the supplemental FTP will reduce exhaust emissions from high-load, “off-cycle” events, and from air conditioning load. Phase in starts in the year 2000.

•

OBD2 system rollouts include misfire detection starting in 1997, catalyst monitoring starting in 1998, and evaporative system leaks starting in 2000.

•

Enhanced evaporative emissions phased in from 1995 to 1998.

•

The ORVR (on-board refueling vapor recovery system) phases in from 1998 through 2000 for passenger cars, and 2001 through 2003 for light-duty trucks.

•

Uncertainties in in-use emissions from computer controlled heavy-duty vehicles when operating at conditions outside the FTP need resolution.

•

New heavy-duty vehicle emission standards come into effect in 2004.

•

The effect of reformulated gasoline and diesel fuels on mass emission rates of HC speciation needs to be taken into account.

•

The in-use effectiveness of I/M programs needs to be monitored and taken into account.

•

The combined effectiveness of OBDII systems and I/M programs needs to be assessed.

•

The impact of mandatory recall compliance needs to be considered.

•

The new regulation of off-road vehicles will affect emissions from these sources.

In the long term, 20 to 50 years, the introduction of new technologies and fuels could have a substantial impact on motor vehicle emissions. A new generation of combustion engine/electric hybrid vehicles with greatly increased fuel economy (2 to 3 times) and reduced emissions (at the ULEV level or lower) is under development with the first production and sales already begun in Japan. A later version could be a fuel-cell hybrid with near zero emissions depending upon the fuel. A continued role for petroleum based motor fuels seems likely as long as supplies remain abundant and prices remain low. Transition to a truly clean fuel, such as pure electric or hydrogen, may not occur until petroleum fuels become in short supply.

CRITICAL ISSUES Overall, it is concluded that the uncertainties in current mobile source emission inventories compromise the confident development of tropospheric ozone control strategies. Improvement of these inventories is critical as they form the foundation of all control strategies.

Specifically: •

At present, large and significant uncertainties exist in the estimates of the mobile source emissions inventory. These uncertainties exist for all vehicle types and classes throughout North America.

•

The mobile source inventory is dynamic. Large changes in the in-use vehicle fleet will continue to occur between now and the year 2010 due to changes in vehicle technology, fuel composition, and vehicle activity.

•

There is currently no routine collection of in-use emissions data from all mobile source types. A process needs to be established to correct this deficiency. For example, validated I/M data should be collected centrally and should be available for use in emission model revisions.

•

The emission inventory needs to include adequate temporal and spatial resolution for ozone modeling. Information on chemical composition of the hydrocarbon emissions also needs to be included. Current models are inadequate in these areas.

•

Accurate activity data (i.e. number of cold starts, hot starts, trip length, soak periods, traffic congestion, day-of-week variations, etc.) is as important as accurate emission rate data. More activity data are required to account for geographic and both shortand long-term temporal changes in driving patterns.

•

A sensitivity analysis of the uncertainties associated with mobile source emissions needs to be included in the emission factor models.

•

Policy makers need to be aware of differences in effectiveness of mobile source control strategies for HC and NOx. For example, both on-and off-road HD diesel emissions are important NOx sources, but are minor sources of HC. Reformulated gasolines are, generally, more effective in reducing HC than NOx.

•

Mobile sources are responsible for about one-half of NOx emissions, more in some areas. These are roughly divided equally among heavy-duty on-road, heavy-duty offroad, and light-duty vehicles. Ozone control strategies that require NOx reduction must pay greater attention to reducing NOx from heavy-duty sources.

•

Pollutant fluxes at the boundaries of modeling domains are important. Rural emissions from on-road and off-road vehicles as well as ships are currently underestimated.

•

Validation studies based on direct measurement of the in-use fleet need to be performed to assess the accuracy of the emissions models. Confidence in the inventory will remain low until agreement is obtained between top-down and bottomup validation approaches.

GLOSSARY AFV AMFA

alternative fuel vehicle Alternative Motor Fuels Act

CARB

California Air Resources Board

CNG

compressed natural gas

CO

carbon monoxide

CO2

carbon dioxide

CRC

Coordinating Research Council

CVS

constant volume sample

DME

dimethyl ether

DOE

Department of Energy

E85

85% ethanol + 15% gasoline fuel (by volume)

EGR

exhaust gas recirculation

EMFAC

CARB mobile source emissions factor model

EPEFE

European Programmes on Emissions, Fuels, and Engine Technologies

FTP

federal test procedure

GVW

gross vehicle weight

HC

hydrocarbon

HD

heavy-duty

HWFET

highway fuel economy test

I/M

inspection and maintenance

IM240

USEPA 240 second chassis dynamometer inspection and maintenance test

LD

light-duty

LDGV

light-duty gasoline vehicles

LEV

low emission vehicle

LPG

liquefied petroleum gas

M85

85% methanol + 15% gasoline fuel (by volume)

MIR

maximum incremental reactivity

MMT

methylcyclopentadienyl manganese tricarbonyl

MOBILE

USEPA mobile source emissions models

MTBE

methyl tertiary butyl ether

MVEI

motor vehicle emissions inventory (CARB)

N2

nitrogen

NLEV

national low emission vehicle

NMHC

non-methane hydrocarbon

NMOC

non-methane organic compound

NMOG

non-methane organic gases

NO x

oxides of nitrogen (NO + NO2)

OBD

on-board diagnostics

OBDII

on-board diagnostic system

ORVR

on-board refueling vapor recovery

PCV

positive crankcase ventilation

PM

particulate matter

RAF

reactivity adjustment factor

RFD

reformulated diesel

RFG

reformulated gasoline

RVP

Reid vapor pressure

SAE

Society of Automotive Engineers

SHED

sealed housing evaporative determination

SO2

sulfur dioxide

TBI

throttle body injection

TCM

transportation control measures

TLEV

transition low emission vehicle

TOG

total organic gas

UDDS

urban dynamometer driving schedule

ULEV

ultra low emission vehicle

USEPA

United States Environmental Protection Agency

VKT

vehicle kilometers traveled

VOC

volatile organic compound

ZEV

zero-emission vehicles

LIST OF TABLES Table 1.

Mobile source vehicles, engines, and fuels. (Listed in approximate order of decreasing use; bold indicates the most important sources of HC and NOx).

Table 2.

North American Motor Vehicle Registrations (AAMA, 1997) (millions).

Table 3.

Emissions from jet engines [gm/kg fuel].

Table 4.

MOBILE5B Emission Rates for the United States (non-California).

Table 5.

Percent of Emissions by Fuel Type.

Table 6.

Exhaust and Evaporative TOG Emissions for Light-Duty Gasoline Vehicles (g/km).

Table 7.

Percent of Passenger Car Emissions Due to Non-Normal Emitters.

Table 8.

Comparison of All-Vehicle Fleetwide Emission Rates with Different North American Emission Models (emission rates in g/km).

Table 9.

Measured On-Road Emissions from Light- and Heavy-Duty Vehicles (mass of pollutant emitted per unit mass of fuel burned).

Table 10.

Mobile Source VOC and NOx Inventories in the United States (1995).

Table 1. Mobile source vehicles, engines, and fuels. (Listed in approximate order of decreasing use; bold indicates the most important sources of HC and NOx). Vehicles light-duty on-road heavy-duty on-road heavy-duty off-road light-duty off-road aircraft ships locomotives

Engines spark ignition compression ignition gas turbine electric steam turbine (marine)

Fuels gasoline diesel jet fuel residual fuel oil liquefied petroleum gas natural gas electricity alcohols

g

Table 2. North American Motor Vehicle Registrations (millions). Canada Passenger cars Commercial vehicles Light-duty trucks Heavy-duty trucks and buses TOTAL

13.2 3.5 na na 16.7

United States 136.1 65.5 56.8 8.7 201.6

Mexico

TOTAL

8.4 3.8 na na 12.2

157.7 72.8 na na 230.5

Table 3. Emissions from jet enginesh [gm/kg fuel]. JT9D-7F, 100% power (take-off) JT9D-7F, 7% power (idle) PW4056, 100% power (take-off) PW4056, 7% power (idle)

g h

HC 0.3 25.0 0.1 0.7

CO 0.5 45.0 0.1 10.0

NOx 46.0 3.0 32.0 5.0

(AAMA, 1997) Both engines are used on Boeing 747 aircraft. The JT9D-7F was certified in 1975, the PW4056, in 1986.

Table 4. MOBILE5B Emission Rates for the United States (non-California). 1997 Vehicle Type Light-Duty Gasoline Vehicles Light-Duty Gasoline Trucks 1 i Light-Duty Gasoline Trucks 2 j Heavy-Duty Gasoline Vehicles Light-Duty Diesel Vehicles Light-Duty Diesel Trucks Heavy-Duty Diesel Vehicles Motorcycles All Vehicles

2010

TOG (g/km) 1.59 1.90

NOx (g/km) 0.96 1.10

TOG (g/km) 1.22 1.44

NOx (g/km) 0.78 0.92

2.58

1.46

2.01

1.30

4.68

3.32

2.41

2.42

0.48 0.69 1.57 2.84 1.83

0.99 1.14 8.51 0.52 1.58

0.32 0.44 1.38 2.82 1.39

0.62 0.76 5.00 0.52 1.24

Table 5. Percent of Emissions by Fuel Type. Year 1997

2010

Vehicle/Fuel Type Gasoline Diesel Motorcycles Gasoline Diesel Motorcycles

TOG 93 6 1 92 6 1

NOx 65 35 0 72 28 0

Table 6. Exhaust and Evaporative TOG Emissions for Light-Duty Gasoline Vehicles (g/km). Year 1997

2010

i j

0-2730 kg GVW 2730-3860 kg GVW

Temperature oC (average, lowhigh) 24, 16-29 35, 22-36 41, 21-41 24, 16-29 35, 22-36 41, 21-41

Exhaust

Evaporative

Total

1.05 1.11 1.17 0.92 0.95 0.97

0.43 0.78 1.69 0.29 0.46 0.93

1.48 1.89 2.86 1.21 1.41 1.90

Table 7. Percent of Passenger Car Emissions Due to Non-Normal Emitters. Year 1997 2010

HC 55% 69%

CO 57% 63%

NOx 27% 34%

Table 8. Comparison of All-Vehicle Fleetwide Emission Rates with Different North American Emission Models (emission rates in gm/km). Year

Pollutan t

1997

TOG NOx TOG NOx

2010

MOBILE5B United States 1.83 1.58 1.39 1.24

MOBILE5 C Canada 1.87 1.40 1.42 1.07

EMFAC7G California

MOBILE5J Juarez

0.86 1.24 0.24 0.56

3.50 0.97 2.07 0.86

Table 10. Mobile Source VOC and NOx Inventories in the United States (1995). Sourc e OnRoad

Fuel Gasolin e

Vehicle Categories

Cars and all gasoline trucks Motorcycles Diesel Diesel trucks and buses Total On-Road OffGasolin Recreational vehicles, Road e utility equipment Diesel Agricultural, construction, commercial boats, trains, and ships Aviatio Aircraft n Fuel Total Off-Road Total On-Road + Off -Road Total, All Sources (also includes area and point sources, excludes biogenic)

VOC (tpd)

NOx (tpd)

15800

15300

101 840 16741 4790

36 5540 20876 360

690

4720

600

400

6080 22821 60000

5480 26356 62500

LIST OF FIGURES Figure 1. North American mobile source fuel consumption, 1995. Figure 2. United States light-duty vehicle emissions standards. Increase in NOx in 1968 was the result of CO and HC control technology; NOx was first controlled in 1976. Crankcase emissions were voluntarily controlled in 1960s. Standards in 2004 are projected default standards of the Clean Air Act Amendments of 1990. Figure 3. Federal Test Procedure (FTP) driving cycle. Test is performed on a chassis dynamometer. Figure 4. United States heavy-duty diesel vehicle emissions standards. Uncontrolled CO, evaporative, and crankcase emissions from diesels are less than those from controlled heavy-duty gasoline engines and are not regulated.

Jet Fuel

Residual Fuel

CANADA

Off-Road Diesel Gasoline On-Road Diesel

Residual Fuel Jet Fuel

US

Off-Road Diesel On-Road Diesel Gasoline

Off-Road Diesel

Jet Fuel

MEXICO

On-Road Diesel Gasoline

Figure 1. North American mobile source fuel consumption, 1995.

PER CENT OF UNCONTROLLED VALUE

160 140 NOx

120 100 80

CO EVAP

60 40

PM

20 HC

0 1960

1970

1980

1990

2000

2010

YEAR

Figure 2. United States light-duty vehicle emissions standards. Increase in NOx in 1968 was the result of CO and HC control technology; NOx was first controlled in 1976. Crankcase emissions were voluntarily controlled in 1960s. Standards in 2004 are projected default standards of the Clean Air Act Amendments of 1990.

SPEED [miles per hour]

60 50 40 30 20 10 0 0

500

1000

1500

TIME [seconds]

Figure 3. Federal Test Procedure (FTP) driving cycle. Test is performed on a chassis dynamometer.

2000

PERCENT OF UNCONTROLLED VALUE

120 CO, EVAPORATIVE, CRANKCASE 100 HC 80 NOX

60

PM

40 20 0 1960

1970

1980

1990

2000

YEAR

Figure 4. United States heavy-duty diesel vehicle emissions standards. Uncontrolled CO, evaporative, and crankcase emissions from diesels are less than those from controlled heavy-duty gasoline engines and are not regulated.

2010

REFERENCES AAMA, 1997. Motor Vehicle Facts and Figures 1997, American Automobile Manufacturers Association. Altshuler, S.L., Arcado, T.D., Lawson, D.R., 1995. Weekday vs. Weekend Ambient Ozone Concentrations: Discussion and Hypotheses with Focus on Northern California. Journal of the Air & Waste Management Association 45, 967-972. AOAQIRP, 1997. Auto/Oil Air Quality Improvement Research Program Final Report, January 1997. BAQMD, 1997. Bay Area Air Quality Management District, Emissions Estimation Manual for the Bay Area ’97 Clean Air Plan, 1997. Barth, M., An, F., Norbeck, J. M., Ross, M., 1996a. Transportation Research Record 1520, TRB, pp. 81-88. Barth, M., Tadi, R., Norbeck, J.M., 1996b. Transportation Congress, 1225-1236. Barth, M.J., Norbeck, J.M., 1993. The Development of an Integrated Transportation/Emissions Model to Predict Mobile Source Emissions, Center for Environmental Research and Technology, University of California, Riverside. Barth, M.J., Norbeck, J.M., 1994. A Power-Demand Approach to Estimating Vehicle Emissions, in Fourth CRC-APRAC On-Road Vehicle Emission Workshop, San Diego, CA, pp. 5-51 - 5-72. Baughcum, S.L., 1996. Subsonic Aircraft Emission Inventories, Chapter 2, Atmospheric Effects of Aviation: First Report of the Subsonic Assessment Project, Thompson, A.M., Friedl, R.R., Wesoky, H.L., (Eds.), NASA Reference Publication 1385. BDM International, 1991. Project Gridlock Report, Technical report prepared and submitted to Ford Motor Company. Beaton, S.P., Bishop, G.A., Stedman, D.H., 1992. Emission Characteristics of Mexico City Vehicles. Journal of the Air & Waste Management Association 42, 1424-1429. Bishop, G.A., McLaren, S.E., Stedman, D.H., Pierson, W.R., Zweidinger, R.B., Ray, W.D., 1996. Method Comparisons of Vehicle Emissions Measurements in the Fort McHenry and Tuscarora Mountain Tunnels. Atmospheric Environment 30, 2307-2316. Bishop, G.A., Starkey, J.R., Ihlenfeldt, A., Williams, W.J., Stedman, D.H., 1989. IR Long-Path Photometry: A Remote Sensing Tool for Automobile Emissions. Analytical Chemistry 61, 671A-677A. Bishop, G.A., Stedman, D.H., Castro, J.D., Davalos, F.J., 1997. On-road remote sensing of vehicle emissions in Mexico. Environmental Science and Technology 31, 3505-3510.

Blake, D.R., Rowland, F.S., 1995. Urban Leakage of Liquefied Petroleum Gas and its Implication on Mexico City Air Quality. Science 269, 953-956. Brooks, D.J., Baldus, S.L., Bandy, W.J., Peltier, R.J., Reuter, R.M., Spark, T.L., 1995. RealWorld Hot Soak Evaporative Emissions - A Pilot Study. SAE Paper 951007. Brownstone, D., Lave, C., 1992. Transportation Energy Use. Universitywide Energy Research Group Report PWP-003, University of California, Berkeley. Burns, V.R., Reuter, R.M., Benson, J.D., Gorse, R.A., Hochhauser, A.M., Koehl, W.J., Painter, L.J., 1992. Effects of Gasoline Composition on Evaporative and Running Loss Emissions. SAE Technical Paper Series, no. 920323. Carter, W.P.L., 1994. Development of ozone reactivity scales for volatile organic compounds. Journal of the Air and Waste Management Association 44, 881-899. Cicero-Fernandez, P., Long. J.R., Winer, A.M., 1997. Effects of Grades and Other Loads on OnRoad Emissions of Hydrocarbons and Carbon Monoxide, Journal of the Air and Waste Management Association 47, 898-904. Dreher, D.B., Harley, R.A., 1998. A Fuel-Based Inventory for Heavy-Duty Diesel Truck Emissions. Journal of the Air & Waste Management Association 48, 352-358. EIA, 1994. Annual Energy Outlook 1994 With Projections to 2010, Energy Information Administration, DOE/EIA-0383(94). EIA, 1996. Alternatives to Traditional Transportation Fuels, 1996 Overview, Energy Information Agency, U.S. Department of Energy. EnvCan, 1997. Conversation with Mr. Leif Stephanson, Vehicle Emissions Section, Environment Canada, September 10, 1997. EPA, 1997a. Regulatory Announcement, Final Emission Standards for Locomotives, EPA420-F97-048, 1997. EPA, 1997b. Technical Highlights, Emission Factors for Locomotive, EPA420-F97-051, 1997. Espinosa, M.E., Paramo, V.H., Sarmiento, J., 1996. Mobile Source Inventories in Mexico. Air and Waste Association 89th Annual Meeting, 96-RA108A.02. FHA, 1986. Federal Highway Administration, 1986. TRANSYT-7F: Traffic Network Study Tool (Version 7F), Washington, DC: U.S. Department of Transportation. FHA, 1988. Federal Highway Administration, 1988. FREFLO User’s Manual, Washington, DC: U.S. Department of Transportation. FHA, 1989. Federal Highway Administration, 1989. TRAF-NETSIM User’s Manual, US Department of Transportation.

FHA, 1992. Federal Highway Administration, 1992. FRESIM - Freeway Simulation Model User Guide, Office of Research and Development. Fujita, E.M., Croes, B.E., Bennett, C.L., Lawson, D.R., Lurmann, F.W., Main, H.H., 1992. Comparison of Emission Inventory and Ambient Concentration Ratios of CO, NMOG, and NOx in California's South Coast Air Basin. Journal of the Air & Waste Management Association 42, 264-276. Fujita, E.M., Lu, Z., Sheetz, L., Harshfield, G., Zielinska, B., 1997. Determination of Mobile Source Emission Source Fraction Using Ambient Field Measurements. Desert Research Institute, Reno, NV. Report to the Coordinating Research Council. Fujita, E.M., Watson, J.G., Chow, J.C., Lu, Z., 1994. Validation of the Chemical Mass Balance Model Applied to Hydrocarbon Source Apportionment in the Southern California Air Quality Study. Environmental Science & Technology 28, 1633-1649. Fujita, E.M., Watson, J.G., Chow, J.C., Magliano, K.L., 1995. Receptor Model and Emissions Inventory Source Apportionments of Nonmethane Organic Gases in California's San Joaquin Valley and San Francisco Bay Area. Atmospheric Environment 29, 3019-3035. Gertler, A.W., Fujita, E.M., Pierson, W.R., Wittorff, D.N., 1996. Apportionment of NMHC Tailpipe vs Non-Tailpipe Emissions in the Fort McHenry and Tuscarora Mountain Tunnels. Atmospheric Environment 30, 2297-2305. Gertler, A.W., Sagebill, J.C., Wittorff, D.N., Pierson, W.R., Dippel, W.A., Freeman, D., Sheetz, L., 1997a. Vehicle Emissions in Five Urban Tunnels. Desert Research Institute, Reno, NV. Report to the Coordinating Research Council, Atlanta, GA. Gertler, A.W., Wittorff, D.N., McLaren, R., Belzer, W., Dann, T., 1997b. Characterization of Vehicle Emissions in Vancouver BC during the 1993 Lower Fraser Valley Oxidants Study. Atmospheric Environment 31, 2107-2112. Grosjean, D., 1992. Atmospheric Environment 26A, 953-963. Guenther, P.L., Stedman, D.H., Bishop, G.A., Beaton, S.P., Bean, J.H., Quine, R.W., 1995. A Hydrocarbon Detector for the Remote Sensing of Vehicle Exhaust Emissions. Review of Scientific Instruments 66, 3024-3029. Harley, R.A., Hannigan, M.P., Cass, G.R., 1992. Respeciation of Organic Gas Emissions and the Detection of Excess Unburned Gasoline in the Atmosphere. Environmental Science & Technology 26, 2395-2408. Harley, R.A., Sawyer, R.F., Milford, J.B., 1997. Updated Photochemical Modeling for California's South Coast Air Basin: Comparison of Chemical Mechanisms and Motor Vehicle Emission Inventories. Environmental Science & Technology 31. Harvey, G.W., Deakin, E., 1993. A Manual of Regional Transportation Modeling for Air Quality Analysis. National Association of Regional Councils, Washington, D.C.

Haskew, H.H., 1997a. Real Time Diurnals. Proceedings of the Seventh CRC On-road Vehicle Emissions Workshop, 2-49, Coordinating Research Council, Atlanta, GA. Haskew, H.H., 1997b. CRC’s Running Loss Test Program - Summer 1997. Proceedings of the Seventh CRC On-road Vehicle Emissions Workshop, 2-59, Coordinating Research Council, Atlanta, GA. Henry, R.C., Lewis, C.W., Collins, J.F., 1994. Vehicle-related hydrocarbon source compositions from ambient data -- the GRACE-SAFER method. Environmental Science & Technology 28, 823-832. Heywood, J.B., 1998. Internal Combustion Engine Fundamentals. McGraw-Hill, New York, p. 915. Hochhauser, A.M., Benson, J.D., Burns, V.R., Gorse, R.A., Koehl, W.J. Painter, L.J., Reuter, R.M., Rutherford, J.A., 1992. Speciation and Calculated Reactivity of Automotive Exhaust Emissions and Their Relation to Fuel Properties. SAE Technical Paper Series, no. 920325. Hoekman, S.K., 1992. Speciated Measurements and Calculated Reactivities of Vehicle Exhaust Emissions from Conventional and Reformulated Gasolines. Environmental Science & Technology 26, 1206-1216. Hublin, M.P. Gadd, P., Hall, D., Schindler, K., 1996. European Progammes on Emissions, Fuels, and Engine Technologies (EPEFE) - Light Duty Diesel Study, SAE Paper 961073. IMP, 1997. Personal communication, Dr. Francisco Guzman, Instituto Mexicano del Petrol. IMT, 1994. Manual Estadistico del Sector Transporte 1993. (1993 Statistic Manual from the Transportation Division). Instituto Mexicano del Transport. INEG, 1997. Personal communication. Instituto Nacional de Estadistica y Geograffa, Mexico. Jack, M.D., 1997. Advances in On-Road Vehicular Emissions Monitoring Instruments and Their Utilization in Real-World Programs for High Emitter and Clean Car Identification. Presented at Seventh CRC On-Road Vehicle Emissions Workshop, San Diego, CA. Jiang, W., Singleton, D.L., McLaren, R., Hedley, M., Dann, T., Wang, D., 1997. Comparison of Organic Compound Compositions in the Emissions Inventory and Ambient Data for the Lower Fraser Valley. Journal of the Air & Waste Management Association 47. Kenski, D.M., Wadden, R.A., Scheff, P.A., Lonneman, W.A., 1995. A Receptor Modeling Approach to VOC Inventory Validation in Five U.S. Cities. ASCE Journal of Environmental Engineering 121, 483-491. Kirchstetter, T.W., Harley, R.A., Kendall, G.R., Hesson, J.M., 1997. Impact of California Phase 2 Reformulated Gasoline on Atmospheric Reactivity of Exhaust and Evaporative Emissions. Paper 97-RP139.01 presented at the 92nd annual meeting of the Air & Waste Management Association, Toronto, Canada.

Kirchstetter, T.W., Singer, B.C., Harley, R.A., Kendall, G.R., Chan, W., 1996. Impact of Oxygenated Gasoline Use on California Light-duty Vehicle Emissions. Environmental Science & Technology 30, 661-670. Koplow, M.D., Jiménez, J.L., Nelson, D.D., Zahniser, M.S., Schmidt, S.E., 1997. Characterization of On-Road Vehicle NO Emissions by Means of a TILDAS Remote Sensing Instrument. Presented at Seventh CRC On-Road Vehicle Emissions Workshop, San Diego, CA. Krieger, R.B., Siewert, R.M., Pinson, J.A., Gallopoulos, N.E., Hilden, D.L., Monroe, D.R., Rask, R.B., Solomon, A.S.P., Zima, P., 1997. Diesel Engines: One Option to Power Future Personal Transportation Vehicles, SAE Paper 972683. Leppard, W.R., Rapp, L.A., Burns, V.R., Gorse, R.A., Knepper, J.C., Koehl, W.J., 1992. Effects of Gasoline Composition on Vehicle Engine-Out and Tailpipe Hydrocarbon Emissions. SAE Technical Paper Series, no. 920329. Lin, C., Milford, J.B., 1994. Decay-adjusted Chemical Mass Balance Receptor Modeling for Volatile Organic Compounds. Atmospheric Environment 28, 3261-3276. Lonneman, W.A., Seila, R.L., Meeks, S.A., 1986. Non-methane organic composition in the Lincoln tunnel. Environmental Science & Technology 20, 790-796. Maldonado, H., 1991. Methodology to Calculate Emission Factors for On-Road Motor Vehicles, Technical Report, California Air Resources Board. Markey, J., 1993. Federal Test Procedure Review Project: Technical Report, #EPA 420-R-93007, Certification Division, Office of Mobile Sources, US EPA. McCracken, J., 1998. EPA looks into makers of dirty diesels. Automotive News, p. 4. McGurrin, M., Wang, P., 1991. An Object-Oriented Traffic Simulation with IVHS Applications, 2nd SAE Vehicle Navigation and Information Systems Conference, Dearborn, MI, 551-561. McLaren, R., Gertler, A.W., Wittorff, D.N., Belzer, W., Dann, T., Singleton, D.L., 1996a. RealWorld Measurements of Exhaust and Evaporative Emissions in the Cassiar Tunnel Predicted by Chemical Mass Balance Modeling. Environmental Science & Technology 30, 3001-3009. McLaren, R., Singleton, D.L., Lai, J.Y.K., Khouw, B., Wu, Z., Niki, H., 1996b. Analysis of Motor Vehicle Sources and their Contribution to Ambient Hydrocarbon Distributions at Urban Sites in Toronto during the Southern Ontario Oxidants Study. Atmospheric Environment 30, 2219-2232. MSTAC, 1997. In Use Deterioration Workgroup Status Report, July 1997, in the Report of the Mobile Source Technical Advisory Subcommittee to the Clean Air Act Advisory Committee. Odum, J.R., Jungkamp, T.P.W., Griffin, R.J., Forstner, H.J.L., Flagan, R.C., Seinfeld, J.H., 1997. Aromatics, Reformulated Gasoline, and Atmospheric Organic Aerosol, Environmental Science & Technology 31, 1890-1897.

ORNL, 1996. Adapted from Davis, S.C. and McFarline, D.N., Transportation Energy Data Book: Edition 16. Oak Ridge National Laboratory, ORNL-6898. PetMex, 1997. Petróleos Mexicanos, Memoraja de Labores 1996. (Mexican Petroleum, Summary of Activities 1996) Unidad de Planeación Corporative de Petróleos Mexicanos. Pierson, W.R., 1996a, Motor Vehicle Inspection and Maintenance Programs - How Effective are They? Atmospheric Environment 30 (21), i-iii. Pierson, W.R., Gertler, A.W., Bradow, R.L., 1990. Comparison of the SCAQS Tunnel Study with other On-Road Vehicle Emission Data. Journal of the Air & Waste Management Association 40, 1495-1504. Pierson, W.R., Gertler, A.W., Robinson, N.F., Sagebiel, J.C., Zielinska, B., Bishop, G.A., Stedman, D.H., Zweidinger, R.B., Ray, W.D., 1996b. Real-world Automotive Emissions Summary of studies in the Fort McHenry and Tuscarora Mountain Tunnels. Atmospheric Environment 30, 2233-2256. Pierson, W.R., Lawson, D.R., Schorran, D.E., Fujita, E.M., Sagebiel, J.C., Tanner, R.L., 1997. Assessment of Non-Tailpipe Hydrocarbon Emissions from Motor Vehicles. CRC Project No. VE-11-7 Final Report, Coordinating Research Council, Atlanta GA. Popp, P., Bishop, G.A., Stedman, D.H., 1997. Development of a High-Speed Ultraviolet Spectrophotometer Capable of Real-Time NO and Aromatic Hydrocarbon Detection in Vehicle Exhaust. Presented at Seventh CRC On-Road Vehicle Emissions Workshop, San Diego, CA. RITNRCC, 1996. Development of Mobile Emissions Factor Model for Ciudad Juarez, Chihuahua, 1996. Radian International LLC for Texas Natural Resource Conservation Commission. Robinson, N.F., Pierson, W.R., Gertler, A.W., Sagebiel, J.C., 1996. Comparison of MOBILE4.1 and MOBILE5 Predictions with Measurements of Vehicle Emission Factors in Fort McHenry and Tuscarora Mountain Tunnels. Atmospheric Environment 30, 2257-2267. Rogak, S.N., Green, S.I., Pott, U., 1997b. Use of Tracer Gas for Direct Calibration of Emission Factor Measurements in a Traffic Tunnel. Journal of the Air &Waste Management Association 47. Rogak, S.N., Pott, U., Dann, T., Wang, D., 1997a. Gaseous Emissions from Vehicles in a Traffic Tunnel in Vancouver, BC. Journal of the Air & Waste Management Association 47. Rykowski, R. A., Darlington, T.L., Heuss, J., 1996. Exhaust Particulate Emissions from Gasoline-Fueled Vehicles. Proceedings of the World Car Conference ‘96, Center for Environmental Research and Technology, University of California, Riverside. Sagebiel, J.C., Gertler, A.W., Dippel, W., 1997. Results of the High Exhaust Emitters Project: Site Characterization, Selection, Feasibility Study. Presented at the Seventh CRC On-Road Vehicle Emissions Workshop, San Diego, CA.

Sagebiel, J.C., Zielinska, B., Pierson, W.R., Gertler, A.W., 1996. Real-world Emissions and Calculated Reactivities of Organic Species from Motor Vehicles. Atmospheric Environment 30, 2287-2296. Schauer, J. J., Rogge, W.F., Hildeman, L.M., Mazurek, M.A., Simoneit, B.R., Cass, G.R., 1996. Atmospheric Environment 30, 3837-3855. Scheff, P.A., Wadden, R.A., Kenski, D.M., Chung, J., 1996. Receptor Model Evaluation of the Southeast Michigan Ozone Study Ambient NMOC Measurements. Journal of the Air & Waste Management Association 46, 1048-1057. Schlappi, M.L., Marshall, R.G., Itamura, I.T., 1993. Truck Travel in the San Francisco Bay Area. Transportation Research Record 1383, 85-94. Signer, M., Hienze, P., Mercogliano, R., Steine, H.J., 1996. European Progammes on Emissions, Fuels, and Engine Technologies (EPEFE) - Heavy Duty Diesel Study, SAE Paper 961074. Sigsby, J.E., Tejada, S., Ray, W., Lang, J.M., Duncan, J.W., 1987. Volatile Organic Compound Emissions from 46 In-Use Passenger Cars. Environmental Science & Technology 21, 466-475. Singer, B.C., Harley, R.A., 1996. A Fuel-Based Motor Vehicle Emission Inventory. Journal of the Air & Waste Management Association 46, 581-593. StatCan, 1997. Quarterly Report on Energy Supply-Demand in Canada, 1996-III, Statistics Canada 57-003-XPB, vol. 21, no 3. Stedman, D.H., Bishop, G.A., Aldrete, P, Slott, R.S., 1997. On-road Evaluation of an Automobile Emission Test Program. Environmental Science and Technology 31, (3), 927-931. Stephens, R.D., Cadle, S.H., 1991. Remote Sensing Measurements of Carbon Monoxide Emissions from On-Road Vehicles. Journal of the Air and Waste Management Association 41, 39-46. Stephens, R.D., Mulawa, P.D., Giles, M.T., Kennedy, K.G., Groblicki, P.J., Cadle, S.H., Knapp, K.T., 1996. An Experimental Evaluation of Remote Sensing-Based Hydrocarbon Measurements: A Comparison to FID Measurements. Journal of the Air & Waste Management Association 46, 148-158. TRB, 1985. Transportation Research Board, 1985. Highway Capacity Manual, Special Report 209, Washington, DC. TRB, 1995. Transportation Research Board. Expanding Metropolitan Highways: Implications for Air Quality and Energy Use, 1995. Special Report 245. National Academy Press, Washington, D.C. Turpin, B.J., Huntzicker, J.J., 1995. Atmospheric Environment 29, 3527-2544. USEPA, 1994. Regulatory Impact Analysis and Regulatory Support Document, Control of Air Pollution; Determination of Significance for Nonroad Sources and Emission Standards for New

Nonroad Compression-Ignition Engines at or Above 37 Kilowatts (50 Horsepower), USEAP, 1994. USEPA, 1996. National Air Pollutant Emission Trends Report 1900-1995, 1996. Office of Air Quality Planning and Standards, USEPA. USEPA, 1997. Appendix D: Modeling and Measuring Induced Demand, 1997. Apogee Research, Inc. C1610-001. Prepared for USEPA, Office of Policy, Planning and Evaluation. USFR, 1994. United States Federal Register, 59 FR 31306, June 17, 1994 Van Aerde, M., 1992. INTEGRATION: A Model for Simulation Integrated Traffic Networks, User’s Guide for Model Version 1.4 d, Transportation Systems Research Group, Department of Civil Engineering, Queens University. Wang, S.C., Paulson, S.E., Grosjean, D., Flagan, R.C., Seinfeld, J.H., 1995. Atmospheric Environment 29, 3527-3254. Warner, C., 1985. Spatial Transportation Modeling, Scientific Geography Series, SAGE Publications, Beverly Hills, California LC# 85-062289. Yarwood, G., Gray, H.A., Ligocki, M.P., Whitten, G.Z., 1994. Evaluation of Ambient Species Profiles, Ambient versus Modeled NMHC:NOx and CO:NOx Ratios, and Source-Receptor Analyses. Systems Applications International, San Rafael, CA. Report to U.S. EPA under contract 68-C1-0059. Zhang, Y., Stedman, D.H., Bishop, G.A., Beaton, S.P., Guenther, P.L., McVey, I.F., 1996. Enhancement of Remote Sensing for Mobile Source Nitric Oxide. Journal of the Air and Waste Management Association 46, 25-29. Zhang, Y., Stedman, D.H., Bishop, G.A., Guenther, P.L., Beaton, S.P., Peterson, J.E., 1993. Onroad Hydrocarbon Remote Sensing in the Denver Area. Environmental Science and Technology 27, 1885-1891. Zhang, Y., Stedman, D.H., Bishop, G.A., Guenther, P.L., Beaton, S.P., 1995. Worldwide OnRoad Vehicle Emissions Study by Remote Sensing. Environmental Science and Technology 29, 2286-2294. Zielinska, B., Fung, K.K., 1994. The Composition and Concentration of Hydrocarbons in the Range of C2 to C 18 Emitted from Motor Vehicles. Science of the Total Environment 147, 281288. Zielinska, B., Sagebiel, J.C., Harshfield, G., Gertler, A.W., Pierson, W.R., 1996. Volatile Organic Compounds up to C20 Emitted from Motor Vehicles, Measurement Methods. Atmospheric Environment 30, 2269-2286.