Scope Study for Expanding the Great Lakes Toxic Emission Regional Inventory to include Estimated Emissions from Mobile Sources

Executive Summary

• Background
• Information Gathering and Survey Results
• Level-of-Detail Recommendations
• Emission Estimation Recommendations
• Framework for Expanding RAPIDS to Include Mobile Sources

Background

The Great Lakes states of Illinois, Indiana, Michigan, Minnesota, New York, Ohio, Pennsylvania, and Wisconsin, in cooperation with the Great Lakes Commission and the U.S. EPA Great Lakes National Program Office are in the process of developing a regional emission inventory for airborne toxic pollutants. This regional emission inventory estimates the emissions of 49 specific toxic air compounds which have been identified as significant contributors to the contamination of the Great Lakes. The first three phases in developing the Great Lakes Regional Air Toxics Emission Inventory have included determination of technical specifications for a regional air toxics database; development of an automated emission inventory software (Regional Air Pollutant Inventory Development System; RAPIDS) with an associated emission factor database and protocols for inventorying hazardous air pollutants; and full implementation of the RAPIDS with its associated protocols to complete an eight-state and one-province emissions inventory for point and area sources.

The Great Waters Program of the U.S. EPA has supported the region's collective efforts to assess the amount of toxic contaminants released in proximity to the Great Lakes since 1993. A new phase of the project was initiated in October 1995 when the Great Waters Program provided additional funding to the Great Lakes Commission to begin work on a regional mobile source toxic air emissions inventory. In January 1996 the Commission contracted with the Minnesota Pollution Control Agency to conduct a comprehensive scoping study to establish the parameters for a regionwide mobile source emissions inventory. Major components of this scoping study include:

1. Identifying the significance of toxic air emissions from mobile sources in the Great Lakes region;
2. Collecting data and developing emission factors for toxic compounds of concern;
3. Determining the level of detail needed to estimate mobile source emissions;
4. Developing techniques for emission estimation and collection and development of emission factors for the toxic compounds of concern;
5. Presenting a framework for the computer software that may be used to estimate mobile source emissions.

Information from this study will be used to develop recommendations for expanding the Great Lakes Region Air Toxics Emission Inventory to include mobile source emissions. If these recommendations are accepted, software will then be developed to expand the RAPIDS. This expansion of the RAPIDS to include mobile sources in the inventory, in addition to area and point sources, will provide a useful tool for a comprehensive estimate of toxic air emissions in the Great Lakes Region.

Information Gathering and Survey Results

The role of mobile sources, principally motor vehicles, in air pollution has been recognized since the 1950s. Mobile sources not only include cars, trucks, and buses, but a variety of off-road vehicles and machinery (examples: planes, boats, all-terrain vehicles, snowmobiles, logging equipment, and lawn and garden equipment).

Emission inventory work has traditionally been focused on the larger industrial, manufacturing, and commercial sources. However, mobile sources have been estimated to contribute significantly to emissions of various pollutants. Using data from 1990-1995 for the Great Lakes region, it has been estimated that mobile sources (including the "transportation" category) contribute approximately 30-40% of the volatile organic compound (VOC) emissions and 10-30% of the particulate matter (PM) emissions.

It can be reasonably concluded that mobile sources are an important source of toxic air pollutants. Mobile sources are an important source of VOC and PM emissions. VOCs and PM emissions include many individual toxic air pollutants. A literature search conducted for this scoping study has identified 375 chemicals and compounds that are emitted by mobile sources.

Concerns have been raised about human health impacts from mobile source emissions. In particular, mobile sources have been estimated to contribute to excess potential cancer risks associated with ambient air concentrations of VOCs and particulate metals. There is also a concern about the potential impact of mobile source emissions on the environment, particularly the Great Lakes and other inland waters, due to emissions of persistent pollutants such as polycyclic aromatic hydrocarbons (PAHs) and particulate metals. This scoping study has further identified 72 chemicals and compounds that are emitted by mobile sources and are of concern to the Great Lakes (Table ES-1). Therefore, the compilation of data on toxic emissions from mobile sources, as well as from point and area sources, in the Great Lakes region is a key component of any strategy to reduce potential direct impacts to human health and indirect impacts through reductions in toxic loadings to lakes.

Due to the concern about potential health and environmental impacts, there is an escalating interest in the emissions data currently available for mobile sources. A survey of the Great Lakes states and the Province of Ontario, Canada, has found that:

• Currently, toxic air pollutants are not included in any mobile source inventories.
• Mobile source emission estimates and inventory work is primarily focused on carbon monoxide (CO), nitrogen oxides (NO_x), and volatile organic compounds. Some limited emissions estimation has included sulfur dioxide (SO₂) and particulate matter less than 10 microns in size (PM10). The pollutants included in these inventories depends upon the classification of the nonattainment area (example: nonattainment for carbon monoxide, or nonattainment for ozone, etc.).
• All Great Lakes states have an emission inventory for nonattainment areas (areas where ambient air concentrations of one or more criteria pollutants exceed a National Ambient Air Quality Standard). The U.S. Environmental Protection Agency (EPA, 1992) guidance is followed to prepare the respective inventories; source categories include highway vehicles, nonroad mobile sources, aircraft, and locomotives.
• Computer models are used to estimate emissions from highway vehicles (MOBILE5a is used by the Great Lakes states to estimate emissions of VOC, NOx, and CO while Ontario uses MOBILE5c; New York uses PART5 to estimate emissions of PM10).
• Emissions are typically calculated for a "typical day" for the "pollutant season". Hourly emissions data for some areas are available from the state of Illinois, while the state of New York and the Province of Ontario can derive daily and hourly emissions estimates if needed.
• Four states update their inventories every three years, four states update their inventories "as needed," and the Province of Ontario prepares an annual inventory.
• Spatial resolution of inventories conducted by the Great Lakes states is based on a "county basis," while the Province of Ontario’s inventory is based on grids (5 x 5 kilometers for urban areas, 50 x 50 kilometers for rural areas, and 100 x 100 kilometers for remote areas).
• Three states have a statewide inventory and Ontario has a province-wide inventory.

Level-of-Detail Recommendations

To determine the level-of-detail for the regional toxic air emission inventory for mobile sources, four factors have been considered:

• Information that was obtained through the emissions inventory survey for the Great Lakes states and the Province of Ontario;
• 15 potential uses of the regional toxic air emission inventory;
• Characteristics of mobile source emissions; and
• Availability of federal and state resources to conduct a mobile sources inventory.

The following recommendations for the mobile sources component of the Great Lakes Region Air Toxics Emission Inventory are based on these four factors.

1. Source Categories. Mobile source categories defined by the EPA (1992) should be used to maintain consistency with EPA emission inventory efforts. The Steering Committee for the Great Lakes Emission Inventory for Toxic Air Contaminants (Steering Committee) has agreed to be consistent with EPA guidance.

a. Highway vehicles

• Light duty gasoline powered vehicles (LDGV)
• Light duty gasoline powered trucks, 0-6,000 pounds (lb) gross vehicle weight (LDGT1)
• Light duty gasoline powered trucks, 6,001-8,500 lb gross vehicle weight (LDGT2)
• Heavy duty gasoline powered vehicles (HDGV)
• Light duty diesel powered vehicles, 0-6,000 lb gross vehicle weight (LDDV)
• Light duty diesel powered trucks (LDDT)
• Heavy duty diesel powered vehicles (HDDV)
• Motorcycles (MC)

b. Nonroad mobile sources (examples)

• Lawn and garden equipment (lawnmowers, snowblowers, trimmers, tillers)
• Industrial equipment (aerial lifts, forklifts, self-propelled elevating platforms)
• Airport service equipment (aircraft and baggage towing tractors, service vehicles)
• Construction equipment (asphalt pavers, rollers, scrapers, rubber-tired dozers)
• Recreational equipment (all-terrain vehicles, off-road motorcycles, snowmobiles)
• Agricultural equipment (tractors, combines, balers, harvesters)
• Recreational marine equipment (inboard and outboard recreational boats)
• Logging equipment (chain saws > 4 horse power, delimbers, log skidders)
• Light commercial equipment (air and gas compressors, welders, pumps)
• Commercial marine vessels (fishing vessels, ocean-going commercial vessels)

c. Aircraft

• Civilian aircraft
• Commercial aircraft
• Military aircraft
• Aircraft auxiliary power units

d. Locomotives (fossil-fuel fired)

• Class I railroads (annual revenues > $93.5 million)
• Class II railroads (annual revenues > $18.7 million, but < $93.5 million)
• Class III railroads (annual revenues < $18.7 million)

2. Pollutants. 72 pollutants are emitted from mobile sources and are of concern to the Great Lakes Table ES-1. The Steering Committee has agreed that emissions of all 72 pollutants should be estimated if possible to provide data to various users of the Regional Air Toxics Emission Inventory.

3. Resolution. The recommendations below represent a compromise between having the "best" inventory possible and the constraints of the existing RAPIDS configuration and the availability of individual state resources (dollars and staff).

1. Spatial. Two options were evaluated: grid-based and county-based. Emissions calculated on a county basis are recommended because most state inventories and the RAPIDS structure are currently set up on a county basis. A grid-based inventory may be an option in the future if and when states update their computer systems and money is available to modify RAPIDS to accept the large volume of grid-based data.

2. Temporal. Four options were evaluated: annual, seasonal, daily, and hourly. Seasonal emission estimates are recommended for compatibility with existing Great Lakes state mobile source emission inventories. It is recognized that hourly or daily estimates would be more accurate due to the fact that mobile source emissions are strongly sensitive to ambient temperatures. However, seasonal emission estimates will allow this temperature-sensitivity to be taken into account by determining a "typical day" or "average day" for each season. In addition, a "hot summer day" and a "cold winter day" should also be considered to more accurately estimate emissions from mobile sources. Seasonal mobile source emission estimates can then be summed to provide an estimate of annual emissions.

   • Spring -- March through May
   • Summer -- June through August
   • Fall -- September through November
   • Winter -- December through February

4. Inventory Update Frequency. Update frequency varies from annual to "as needed" in the Great Lakes region. Every three years is recommended for an actual inventory due to the complexity of estimating mobile source emissions, with annual updates provided by using growth factors (for the years when emission data are not collected). Annual updates of the mobile sources emission inventory will coincide with the annual inventory for point and area sources to provide a more complete picture of emissions in the Great Lakes region.

Emission Estimation Recommendations

1. Highway Vehicles. Three options were evaluated: 1) using toxic emission factors in combination with activity levels (grams of pollutant emitted per mile for a specific vehicle type with a specific control device using a specific fuel and operating on a specific road type, multiplied by an activity level which is expressed as "vehicle miles traveled"); 2) using toxic emission factors in combination with fuel usage (grams of pollutant emitted per unit of fuel burned, combined with fuel sales data); and 3) estimating total organic gas (TOG; includes hydrocarbons, aldehydes, alcohols, and other oxygenated compounds) and PM emissions using available emission factor computer models and then speciating the TOG and PM emissions into the respective toxic air pollutant emissions for each source category.

Option #3 is recommended for use in the mobile sources inventory due to limited emission factor data currently available and the uncertainty as to when emission factors will be developed for the numerous combinations of vehicle types, control devices, fuels, and roads.

Under Option #3, TOG and PM emissions for highway vehicles can be estimated with existing EPA emission factor computer models (MOBILE5a, MOBIL5c for TOG; PART5 for particles 1-10 microns in size). These computer models account for the effects of numerous vehicle parameters on the amounts of pollutants emitted.

Travel activity data ("vehicle miles traveled"; VMT) can be obtained from several information sources: 1) the U.S. Department of Transportation, Federal Highway Administration; 2) travel demand network models; and 3) individual state vehicle and travel monitoring systems. Typically, the state data is more consistent and statistically robust than the federal data, and more reliable than the model-generated data. It is further recommended that where state VMT data is available, the state data should be used.

The estimation process for the TOG and PM emissions can be represented by a simple equation:

The TOG and PM emissions can then be used to estimate the emissions of specific toxic air pollutants by using available TOG and PM speciation profiles for mobile sources. It is recognized that limited data have been used to develop some of the speciation profiles; that some toxic air pollutants are not included in the profiles; and that overall there are a limited number of profiles available for use. In addition, EPA has cautioned against using existing PM speciation profiles to estimate mobile source emissions of individual toxic air pollutants. However, the recommended approach of estimating TOG and PM emissions and then using available speciation profiles to estimate individual emissions of toxic air pollutants is currently the most feasible approach for the Great Lakes states and the Province of Ontario in preparing an inventory for highway vehicles.

2. Nonroad Mobile Sources. Nonroad mobile sources have been divided into 10 subcategories. In addition, there are 79 equipment types within the 10 subcategories. Each of the 79 equipment types can consist of three possible engine types (diesel, 4-stroke gasoline, 2 stroke gasoline with substitution by propane or compressed natural gas) that must be considered in the emissions estimation process. Limited data is available to estimate emissions for most subcategories of nonroad mobile sources. A two-step process is recommended to calculate county-level emissions for each nonroad mobile source subcategory:

• Step 1. When emission factors for specific toxic air pollutants are available: Emission factor is multiplied by activity data for each nonroad mobile source subcategory, and for each equipment type within each subcategory.

  Exhaust & Crankcase Emissions:

  Toxic Pollutant Emissions (g/yr)

  =

  Emission Factor (g/hp-hr) (g/gal)

  x

  Equipment Population (number)

  x

  Annual Hours Used (hrs/yr)

  x

  (Av. HP x LF )   (hp)

  Where:	Av. HP	=	average horsepower
  	g/yr	=	grams per year
  	g/hp-hr	=	grams per horsepower per hour
  	g/gal	=	grams per gallon of fuel

  Evaporative Emissions:

  Toxic Pollutant Emissions (g/yr)

  =

  Emission Factor (g/hr) (g/day)

  x

  Equipment Population (number)

  x

  Evaporative Emission Season (days/yr)

  Where:	g/yr	=	grams per year
  	g/hr	=	grams per horsepower per hour
  	g/day	=	grams per day
  	Evaporative emission season: ~ 229 days per year for VOCs

  Emission factors developed for larger combustion engines (diesel, gas, propane, compressed natural gas) for specific toxic air pollutants can be applied to small 2-stroke and 4-stroke engines. Data is also available from the Society of Automotive Engineers for developing emission factors. In addition, the State of Wisconsin has developed an air toxics inventory for nonroad engines and this inventory can be used as a starting point for the other Great Lakes states and the Province of Ontario when estimating nonroad mobile source emissions.

• Step 2. When emission factors are not available for specific toxic air pollutants: a) VOC and PM emissions are estimated; b)VOC emissions are then converted to TOG emissions; c) speciation profiles are used to estimate individual toxic air pollutant emissions from the TOG and PM emissions.

  Two techniques are available to estimate VOC and PM emissions in Step 2:

  • The first technique is to use EPA inventories for VOC and PM emissions that are available for 33 specific areas. These VOC and PM emission inventories are considered to be geographically representative of areas with significant pollution problems. These existing EPA inventories do not include commercial and marine vessels. These inventories can be adjusted to a specific area in a state by the following formula:

    County Emissions

    =

    County Population EPA Area Population

    x

    EPA Area Emissions

  • The second technique is to use VOC and PM emission factors in combination with activity data. The equations shown in Step 1 above for estimating exhaust and crankcase emissions and evaporative emissions also can be used to estimate VOC and PM emissions in Step 2.

  In both Steps 1 and 2, equipment population in a county is a key unknown activity variable. Other activity parameters, such as annual hours of use, average horsepower, and load factor, remain essentially constant in a specific area as well as at a state and national level. While actual survey data on equipment populations is limited at best, activity indicators can be used to estimate equipment populations. Regression analysis using human population data and activity indicator data can provide reasonable estimates of equipment populations. States can use their own activity indicator data, develop activity indicators using data from the Power System Research database, or rely on EPA activity indicators and adjust the EPA data to their specific areas of interest using the following formula:

  County Equip. Pop.

  =

  County Activity Indicator National Activity Indicator

  x

  National Equip. Pop.

3. Aircraft. Two options were considered for this category: 1) use emission factors in combination with activity level data to estimate individual toxic air pollutant emissions; and 2) estimate TOG and PM emissions and then use speciation profiles to estimate individual toxic air pollutant emissions. Due to the lack of emission factors for aircraft, the only feasible option for estimating aircraft emissions is Option 2.

The general steps for estimating emissions are the same for all categories of aircraft and include:

• identifying all airports in each county;
• determining the mixing height to be applied to all landing and take-off cycles;
• defining the fleet make-up for each aircraft category using each airport;
• determining the airport activity as the number of landing and take-off cycles for each aircraft category;
• selecting hydrocarbon (HC) and PM emission factors for each operation mode for each engine model for each aircraft model;
• estimating the length of time that each aircraft model spends in each of the 5 modes of the landing and take-off cycle at each airport;
• calculating HC and PM emissions based on the airport activity, the length of time each aircraft model spends in each mode of the landing and take-off cycle, and aircraft emission factors for each model of engine in each county;
• aggregating the HC and PM emissions for each engine type and each aircraft category;
• converting HC emissions to TOG emissions; and
• speciating the TOG and PM emissions into the individual toxic air pollutant emissions.

Critical information that needs to be adjusted in estimating aircraft emissions is the length of time each combination of aircraft category-aircraft model-engine type-engine model is in each mode of the landing and take-off cycle.

4. Locomotives. Two potential approaches were considered: 1) use emission factors in combination with activity level data to estimate individual toxic air pollutant emissions; and 2) estimate TOG and PM emissions and then use speciation profiles to estimate individual toxic air pollutant emissions. The analysis shows that a two-step estimation process can be used to estimate locomotive emissions for line haul operations (travel between two distant locations) and yard (railcar switching) operations.

• Step 1. Emission factors for large diesel engines can be used to estimate emissions from locomotives. These emission factors can be obtained from the FIRE 5.1a database. Data obtained from literature searches on diesel combustion engines can also be used to calculate emission factors.

  Activity level data for locomotives is typically fuel consumption (gallons used) or heat input rating which is reported as million British thermal units (MM Btu). The heat input rating can be derived by multiplying fuel consumption by the heating value of diesel fuel. Fuel consumption data can be obtained from operating records that Class I railroad companies are required to keep. Total system-wide fuel consumption for a Class I company can be partitioned to the county level for estimating county-level emissions. For Class II and III railroad companies, fuel consumption data can be estimated from track length and track density data for each respective company. Again, system-wide fuel consumption data can be partitioned to the county level for estimating county-level emissions.

• Step 2. If emission factors are not available, then TOG and PM emissions must be estimated for the locomotive subcategory, and then speciation profiles used to partition the TOG and PM emissions into individual toxic air pollutant emissions. Important information needed in estimating emissions in Step 2 includes: fuel consumption, HC and PM emission factors (EPA, 1991), a factor to convert HC emissions to VOC emissions and then convert VOC emissions to TOG emissions. Speciation profiles can then be used to partition the TOG and PM emissions into individual toxic air pollutant emission estimates.

  The basic equation for estimating emissions (line hauling or yard operations) is as follows:

  Toxic Pollutant Emissions (lb/yr)

  =

  Fuel Consumed (gallons)

  x

  EF-HC (lb/yd-lc) or EF-PM (lb/gal)

  x

  Conversion Factor (HC to TOG)

  x

  %W

  Where:	EF-HC	=	Emission factor for hydrocarbons (lb/yd-lc or lb/gal)
  	EF-PM	=	Emission factor for particulate matter (lb/yd-lc or lb/gal)
  	%W	=	Weight percent of the toxic pollutant in the TOG or PM emissions
  	lb/yd-lc	=	Pounds of emissions per yard locomotive
  	lb/gal	=	Pounds of emissions per gallon of fuel consumed

  For yard operations, the number of locomotives involved in railcar switching operations can be used instead of fuel consumption data. The number of locomotives being used in yard operations can be obtained from visual observations of yard activities or through surveys of railyard managers.

  Emission factors for HC and PM differ for line hauling and yard operations. These emission factors are derived from national average railroad fleet and operations data. This data may differ significantly from data that may be available at the state or local level. To provide the best estimate of HC and PM emissions, state and local data should be used whenever it is available.

Framework for Expanding RAPIDS to Include Mobile Sources

To include mobile source emission estimates in RAPIDS, the following tasks will need to be completed:

1. For highway vehicles, establish a link between RAPIDS and the TOG and PM emission factors provided by EPA models (MOBILE5a, PART5). These emission factors need to be included in RAPIDS because they are specific in road functional class differ from county to county, and from season to season. New activity parameters will need to be added to RAPIDS to match the activity data for highway vehicles.
2. For nonroad mobile sources, expand the RAPIDS reference tables to include national nonroad equipment populations and national activity indicators. This information will then allow county level equipment populations to be estimated using a linear relationship between equipment population and activity indicators.
3. For aircraft, establish a link between RAPIDS and the Federal Aircraft Administration’s Aircraft Engine Emission Database (FAEED) and Program. The FAEED database contains emission factors for various aircraft engines and data correlating engines to specific aircraft. The FAEED program is capable of computing the total emissions of HC and PM produced by a specific fleet. With the linkage, RAPIDS can estimate the county level HC and PM emissions. The factors to convert HC emissions to TOG emissions will also need to be added to RAPIDS.
4. For all categories of mobile sources, incorporate representative emission factors and speciation profiles into RAPIDS. Important areas to focus efforts include:
   • "compounds of concern" to the Great Lakes that are not included in existing EPA speciation profiles;
   • emission factors and speciation profiles for diesel vehicles and engines;
   • emission factors and PM speciation profiles for individual metals and their respective compounds, other than lead.
5. For estimating mobile source emissions on an annual basis, include within RAPIDS appropriate growth indicators (population data, fuel consumption data, etc.) so that growth factors can be calculated and subsequently used with corresponding algorithms to project emissions for those years when emission data are not collected.
6. To collaboratively and effectively handle mobile source data as well as point and area source data, modify and enhance the Import/Export, Quality Control Checker, Reports, and Estimator functions in RAPIDS.

Table ES-1. Pollutants of concern emitted from mobile sources