There is no specific requirement for each state to report each facility or area source by identifying its source, device, and process; this information will be handled by using a consistent set of process codes (SCC and AMS codes), and SIC codes for both facility and area sources. The SCC/AMS codes and SIC codes that will be used to upload source/device/ process data into the regional repository of emissions data and estimates are included in the RAPIDS reference tables (see Section 7.1).

In RAPIDS, control devices are treated as separate devices, not as part of the device that created the emissions. This approach supports a more accurate representation of the physical connections between devices that create emissions, control emissions, and release emissions to the atmosphere. Data uploaded into the regional repository must be able to generate information that associates control efficiency with a specific device having its own device ID number and device code.

4.2 Specific EETs

The following discussion describes the available EETs and their prioritization for use.

4.2.1 Available EETs

The EETs available for use in the regional inventory development effort are:

• Mass balance, which can consist of either measured or calculated inputs and outputs;
• Emission factors, which can be either source-specific or generic;
• Speciation of particulate matter (PM) or total organic gases (TOG), which can be either generic or source-specific;
• Engineering calculations;
• Direct measurement; and
• Process simulation software.

Each of the categories of EETs listed above, plus the EET for on-road motor vehicles, is discussed in Section 4.4.

The first four categories of EETs listed above are generally applicable to both facility sources and area sources, while direct measurement and process simulation software are applicable only to facility sources.

In addition, source-specific emission factors and source-specific speciation profiles can be generated from the following data:

• Material Safety Data Sheets (MSDSs);
• Fuel analysis;
• Source tests; and
• Laboratory analysis.

The complete list of possible EETs and their associated method codes, which are to be included with each emission calculation entered into the regional repository, are shown in Table 4.2.

4.2.2 Specific Emission Estimation Requirements

The most appropriate EET for every occurrence of a source/device/process cannot be predetermined. Even when presented with all available information on two or more competing EETs for a specific source/device/process, determining the most accurate EET may not be possible. For example, one cannot tell which of the following EETs applicable to a given source/device/process would provide the best emissions estimate:

• Using an emission factor with a B rating;
• Using an engineering calculation that relies on default values because of the absence of measured values;
• Using a source-specific specie profile provided by a source operator that is a composite of similar materials used throughout a facility; and
• Using an source-specific emission factor based on a source test applicable to a single point in time conducted for a compliance inspection.

However, in most cases certain EETs are known to be the most accurate in nearly all occurrences of a source/device/process. The available EETs described previously have been evaluated for use in the regional inventory to ensure that only the most appropriate EETs are used and that consistent methods are used from state to state. Listed below in order of preference (from most preferred to least preferred) are the EETs considered to be appropriate (lacking any device/process specific information):

1. Continuous emission monitor judged to meet acceptance criteria for such devices and which does not have any significant gaps in recorded data.
2. Process simulation software judged to be able to replicate the emission measurements on which its simulation is based.
3. Engineering calculations judged to be applicable to the device/process.
4. Mass balance using measurements of input and output streams judged to be based on measurements that can be relied on to generate an emission rate at least as accurate as the next most accurate technique.
5. Mass balance using calculated inputs and/or outputs judged to be representative of the device/process and at least as accurate as the next most accurate technique.
6. Source-specific emission factors judged to be representative of the device/ process and at least as accurate as the generic emission factor, if any.
7. Generic emission factor judged to be representative of the device/process.
8. Source-specific speciation factors used in conjunction with emissions estimated using another EET where the factors are judged to be representative of the device/process.
9. Generic speciation profiles used in conjunction with emissions estimated using another EET where the factors have a rating of C or better and which have been reviewed and accepted by the facility operator.

Any EET not included in the above list is considered to be unacceptable for use in developing the regional inventory.

The available EETs were further evaluated as they apply to specific source/device/processes. The conclusions were:

• EETs 1, 2 and 3 as described above are always the most preferable in the order listed;
• EET 9 is always the least preferable; and
• The applicability of EETs 4, 5, 6, 7, and 8 needs to be considered in the context of the source/device/process in which they may be used.

Table 4-3 shows the applicability of the mass balance (4 and 5), emission factor (6 and 7) and source-specific speciation (8) to specific source/device/processes defined by SCC-AMS codes. The priority of use where more than one EET has been identified is also shown. The EETs identified in this table are:

• MASS BALANCE: Mass balance (either measured or calculated);
• EMIS FACTOR: Emission factors (either generic or source-specific, with source-specific always having the higher priority);
• SPECIATION: Source-specific speciation;
• TANKS2: U.S. EPA TANKS Model version 2;
• SIMS: U.S. EPA Surface Impoundment Model System;
• CHEMDAT7: U.S. EPA spreadsheet for treatment, storage, and disposal facility (TSDF) processes;
• LAEEM: The Landfill Air Emissions and Estimation Model;
• MANUAL: Any unspecified engineering calculation that must be applied in lieu of any other EET.

Each state must use the priorities discussed above in determining which EET to apply to a device/process.

4.3 Overall Inventory Development

This section discusses the steps required to develop the state portion of the regional inventory, plus those requirements that are common to all EETs. The following discussion describes each step that a state needs to take to identify the EET applicable to a source/device/process, and to provide that data in a form appropriate for inclusion in theregional inventory. (Each step needed to carry out specific EETs is discussed in Section 4.4.)

These steps assume certain facts regarding how each state manages its emissions data. Specifically, it is assumed that a database of some sort (not necessarily RAPIDS) currently exists for the management of facility sources of criteria air pollutants (volatile organic compounds [VOC], nitrogen oxides [NO_x], sulfur oxides [SO_x], PM, carbon monoxide [CO] and lead). It is also assumed that the procedures used to develop the facility source and area source criteria pollutant emissions estimates generally follow U.S. EPA guidance regarding the development of such inventories. Additional considerations of inventory development are described below.

4.3.1 Facility and Area Source Reconciliation

As described previously, processes associated with facilities/devices are represented by an eight-digit SCC, while those related to area and mobile sources are denoted by a ten-digit AMS code. A given category of emissions can be comprised of both facility and area sources. For example, perchloroethylene dry cleaning has an SCC code of 4-01-001-01 and an AMS code of 14-20-xxx-055, depending on the volume of dry cleaning. This leads to the possibility that the same device/process will be double-counted. For example, if perchloroethylene emissions from all dry cleaners located in a given county, derived from a per capita emission factor totalled 25 tons/year and a relatively large dry cleaner was inventoried as a discrete facility with perchloroethylene emissions of 4 tons/year, the correct area source emission estimate for the county would be 21 tons/year (i.e., 25 - 4 = 21 tons/year). To eliminate the double counting of emissions, a reference table relating AMS codes to SCC codes for the same source/device should be developed by the states.

Each state must ensure that facility and area source emissions are reconciled to eliminate double counting. This involves reviewing and comparing facility source and area source emission estimates for the same type of source/device/process and then calculating the appropriate values of the area source activity and area source emissions that belong to the area source category. The area source activity is thus calculated by either of the following, as appropriate:

1. If the activity data for the facility and area sources are on the same basis (e.g., gallons of solvent), then the area source activity is the difference between the total activity and the facility source activity.
2. If the activity data are on a different basis (e.g., gallons of solvent for the facility source and population for the area source), then the area source activity is calculated indirectly by first calculating the area source emissions as the difference between total and facility emissions and then back-calculating the activity that would produce the previously determined area source emissions.

4.3.2 Use of Information in Existing State Databases and AIRS

Facility source data are required for sources that emit certain amounts of criteria emittants for various air quality planning and regulatory purposes. These data are generally stored in a database designed for that purpose. In addition, these criteria pollutant data are transmitted to the U.S. EPA, where some or all of the state's emissions and estimates data are stored in AIRS. These databases may be an important source of information that can be used in the regional air toxics emissions inventory development effort.

To be usable for the purpose of preparing/compiling the regional air toxics emissions inventory, each state will need to develop a method to use the data to generate emission estimates for the target compounds. These methods can include extracting data from the facility source database and manually calculating the emissions, or using the database itself to do the calculations. Whatever method is chosen, the state will need to review the methods and associated activity data residing in its facility source database to determine that they are consistent with those required by this Protocol.

4.3.3 Incorporating Emissions Control

The regional database (RAPIDS) uses specific metrics or attributes to enter information on the effect of control devices. These metrics, which must be used when uploading data on these devices to the regional inventory, are:

• CNTL EFFIC: Control efficiency (%) of an emittant;
• RULE EFFECT: Rule effectiveness (%) of an emittant;
• RULE PENET: Rule penetration (%) of an emittant; and
• CAPT EFFIC: Capture efficiency (%) of an emittant.

The relation of these metrics to controlled and uncontrolled emissions is:

CEmA=UEm [(1 - (CNTL EFFIC/100)(RULE EFFECT/100)(RULE
  PENET/100)(CAPT EFFIC]

where:

CEm_{A = Controlled emissions of emittant A; and
UEmA = Uncontrolled emissions of emittant A.}

Control efficiency is the relative amount of control achieved by a control device or some other means of reducing emissions from a specific pollutant. Default values are available for various types of control equipment. These values should be used only when actual data are unavailable and the use of the default is documented.

Rule effectiveness is an adjustment to the control efficiency that reflects the actual performance of the controls in practice. Control equipment performance may be adversely affected by age of the equipment, lack of maintenance, or improper use. A default value of 0.80 for rule effectiveness has been recommended by U.S. EPA lacking other information.

Rule penetration is a measure of the overall extent to which a given regulation actually covers a given source/device/process in a given area source category (rule effectiveness is not used for facility sources). For example, regulations on underground gasoline tank filling may apply only to stations above a specified size cutoff, or to facilities built after a certain date.

                    Uncontrolled emissions
                     covered by regulation
    Rule penet=  ________________________     x  100
                    Total uncontrolled emissions     

For example, if a rule only affects sources built since 1987 and 20% of the facilities fall into that category, then rule penetration is equal to 20%, assuming that there is no difference in average uncontrolled emissions of facililties built before and after 1987. Default values are not feasible for rule penetration because it is highly category- and location-dependent.

Capture efficiency is the relative amount of an exhaust stream that reaches a control device (i.e., the amount of the stream that does not leak or otherwise escape into the atmosphere before reaching the control device).

The following defaults should be used in the absence of any value specified for the parameters listed directly above, or when the emission factor being used includes the effects of control equipment:

• CNTL EFFIC: 0%
• RULE EFFECT: 100%
• CAPT EFFIC: 100%
• RULE PENET: 100%

4.3.4 Activity and Emission Units

All emissions data entered into the regional repository are to be in units of actual pounds emitted in the specified calendar year. All other data may be entered in any convenient units. In all cases, the units used must also be entered.

4.3.5 Scale-Up for Missing Sources

In some situations, only a fraction of the sources in a specific source type can be identified, and the required data collected. This may occur for either facility source emissions or area source emissions. When this happens, the amount of activity in the missing fraction may be estimated by "scaling up" the activity found in the known fraction. If the device/process uses both the facility and area source approaches, the missing fraction is incorporated into the activity data used for the area source; if only the facility approach is used, an area source will need to be added to be able to add in the missing fraction. Typically, employment data are used as an indicator to "scale up" the inventory to account collectively for missing sources and emissions for an area source inventory. Parameters other than employment, such as sales data or number of facilities, can be used to develop emissions estimates, if deemed more appropriate and the data needed to perform this adjustment are readily available. However, employment data is generally more readily available than other types of data that might be used to scale-up emission estimates. The steps taken to scale-up emissions must be fully documented.

4.3.6 Work Breakdown of Tasks Common to All EET Categories

Table 4-4 identifies each of those tasks that have been identified as necessary for a state to develop data for upload into the regional repository and which are applicable to all EETs. This table is organized in work breakdown structure format. The main task (e.g., GLC Toxic Emissions Data Development) is broken down into the subtasks needed to complete that task. Each subtask and each level below subtask are further broken down as necessary to fully identify the work required. Similar work breakdown structure tables are included in the discussion of each EET category.

4.4 General Discussion of Each EET

This section contains a general discussion of the EETs that are available for estimating emissions for facility sources and area sources. The EETs included are:

• Mass balance;
• Source-specific and FIRE emission factors;
• Speciation of VOC and PM;
• Extrapolation;
• Engineering calculations;
• Direct measurement;
• Process simulation software; and
• On-road motor vehicle methods.

In some cases, more detailed information is needed than can be presented in a discussion of an EET. In these cases, the user of this Protocol is referred to the Bibliography in Section 8.1.

4.4.1 Mass Balance

Mass balance is a method that estimates emissions by analyzing inputs of a material to a process minus consumption, accumulation, and loss of that material during a process. Mass balance calculations must account for all routes of process material inflow and outflow, as well as any accumulation or depletion of the material in the equipment (i.e., devices), including control devices, and as a result of any chemical reactions associated with the process. Mass balance calculations can be applied in a variety of different ways. A mass balance calculation can be made for a specific device/process at a facility source or for an area source. National usage data for a material that is disaggregated to a state or county level is a form of a mass balance calculation.

Mass balance calculations for a device/process at a facility source can be based on either measurements of input and output streams or calculated inputs and/or outputs. In the former, inputs and outputs are directly measured, and the difference is calculated. In the latter, the inputs, outputs, and other terms are calculated from other parameters (e.g., transfer efficiency, source-specific speciation profiles). In either case, the appropriateness of using this technique must be judged on its ability to accurately calculate a difference between the various terms of the mass balance equation.

The key to developing useful data used for mass balance calculations is to determine how accurate the activity data must be to support this calculation. For facility sources, the required activity data can only be obtained from the facility operator. For area sources, activity data applicable to all source/device/processes covered in a given region must be properly aggregated. Table 4-5 describes the work breakdown for mass balance.

Section 4.8.1 presents an example of a calculation using mass balance.

4.4.2 Source-Specific and FIRE Emission Factors

Emission factors can be used to estimate emissions of the target compounds from a wide range of sources, including facility and area sources. An emission factor expresses emissions as a ratio of the amount of an emittant released to a process-related parameter or measurement, frequently expressed as the amount of pollutant emitted per throughput of a process or piece of equipment (i.e., device), or per quantity produced or processed (for example, pounds of a particular substance emitted per pounds of product produced). Examples include the pounds (lbs) of NO_x per gallon of residual oil burned, and the pounds of VOC per year of chemical process unit operation.

The general equation for calculating uncontrolled emissions using an emission factor is:

Em_A = EF_A C_A2 ... A₁ A₂ ...

where:

EmA = Emissions of emittant A;

EF_A = Emission factor of emittant A;

C_AA = 0 or more conversion factors; and

A=1 of more activity values.

Activity data (A) quantify the activity associated with a given emission factor (examples include capacity of a storage tank, amount of fuel burned, and hours of usage of a device). The conversion factors (C_A) are those factors needed to apply the emission factor to the activity data available. This includes factors such as heat content of fuel and various efficiency factors.

The following are examples of the use of this equation showing the parameters included with their common units.

• Calculation for fuel combustion requiring a single conversion factor and a single activity value:

  EMA = EF (# benzene/MMBtu heat) x C (MMBtu heat/ton coal) x A (ton coal)

• Calculation of dust from hauling dirt requires two activity values:

  EMA = EF (# PM/yard³-mile dirt hauled) x A1 (yard³ dirt) x A2 (miles hauled)

Table 4-6 describes the work breakdown for generic emission factors.

4.4.2.1 Generic Emission Factors

The generic emission factors to be used for developing the regional inventory are contained in FIRE/GLC, a database/software system containing U.S. EPA-approved emission factors. The majority of the emission estimates made for the regional inventory are expected to be made using emission factors obtained from FIRE/GLC (which is discussed further in Section 7.0).

The FIRE/GLC contains a number of emission factors that cannot be used directly because the emission factor is:

• Associated with a partial SCC code;
• Expressed as a range;
• Less than the value listed; or
• Expressed as a function.

In these cases, the appropriate emission factor must be determined by a review of the reference data contained in FIRE/GLC to determine if the factor is applicable to the given device/process.

In addition, many of the emission factors in FIRE/GLC are controlled; i.e., the emission factor is applicable to the emissions after they have passed through one or two control devices. The control devices are identified in FIRE/GLC. In order to use the controlled emission factors, the subject device/process must have the same configuration of control devices.

4.4.2.2 Source-Specific Emission Factors

Source-specific emission factor data are similar to and are used in the same manner as generic emission factor data, except that they are applicable to a specific source/device/process. Source-specific emission factors can be developed from fuel analyses, source tests, and laboratory analysis. These data are often available as a part of the information developed for permitting and enforcement purposes; sometimes the facility operator will have the required data as a result of monitoring and analysis performed by the facility operator for purposes of quality control and process optimization. Inquiries of the facility operator are necessary to determine the existence and extent of the data available for use to develop emission factors for a specific source/device/process.

Source-specific emission factors may also be available in the form of emissions per hour that a process creates at a device. These data are often developed as a result of permit processing and may be used as an allowable emission rate for the device/process. Emission rate data can be used if it is representative of the actual operating conditions and is not simply an upper limit that is seldom achieved.

Section 4.8.2 presents examples of emission calculations using emission factors.

4.4.3 Speciation of PM or VOC using a Generic or Source-Specific Profile

Speciation profiles are sets of tabulated data that relate the amount of a specific material as a fraction of the amount contained in an aggregation of materials. Specifically, TOGs are composed of many hundreds of specific organic compounds. Likewise, PM is composed of numerous specific elements and compounds that may be present at different particle sizes. Speciation profiles provide a means of estimating the amount of emissions of a target compound included as a specie of either TOG or PM. This method is applicable to facility and area sources.

4.4.3.1 Generic Speciation Profiles

Generic speciation profiles were developed to speciate emissions for input to photochemical models, not for deriving air toxic emission estimates. Therefore, a great deal of uncertainty exists in an emission estimate for one of the target compounds derived using a speciation profile. However, in many instances, there may be no other means of estimating emissions of one or more of the target compounds from a given source. In those instances, the state may choose some estimate of the emissions rather than no estimate at all, providedthe generic profile has a quality rating of C or better and (for facility sources). It is further recommended that the resulting emissions estimate be reviewed by the facility operator. At a minimum, estimates of target compound emissions derived from speciation profiles will help in prioritizing emission factors/EET for future development.

The U.S. EPA's SPECIATE database (U.S. EPA, 1988) contains the species profiles for a variety of source types (organized by profile number). Figure 4-3 is an excerpt from the SPECIATE database shows the percent weight of three organic compounds found in emissions from fluorocarbon manufacturing. SPECIATE (as well as RAPIDS) contains a list of those SCC-AMS codes considered applicable to a given profile.

Criteria pollutant inventories are sometimes developed as VOC; the profiles in SPECIATE are defined in terms of TOG. To use the SPECIATE profiles, VOC first needs to be converted to TOG using the following equation:

EM_TOG= Em_VOC * EFact_x/EFact_x - Fact_Meth - Fact_Eth

Em_{TOG = TOG emissions;
EmVOC = VOC emissions;
Factx = Speciation profile factor for specie x;
FactMeth = Speciation profile factor for methane; and
FactEth = Speciation profile factor for ethane.}

In addition, corrections must also be made to adjust certain categories for formaldhyde to measurement error.

To calculate emissions for a target compound using SPECIATE data, the following equation must be used:

Em_A= Em_TOG/PM Fact_A

where:

Em_{A = Emissions of emittant A;
EmTOG/PM = Emissions of either TOG or PM; and
FactA = The profile factor in SPECIATE for emittant A for either TOG or PM, as appropriate.}

Table 4-7 describes the work breakdown for generic speciation.

4.4.3.2 Source-Specific Speciation

Source-specific speciation data are similar to and are used in the same manner as generic speciation data, except that they are applicable to a specific source/device/process. As such, these data must be obtained from the facility operator. Typically these data are obtained by an analysis of materials used or generated at the facility, or using information sources such as MSDS, fuel analyses, source tests, and laboratory analysis. Often these data are routinely available as a result of monitoring and analysis performed by the facility operator for purposes of quality control and process optimization. Inquiries of the facility operator are necessary to determine the existence and extent of the data available for use to speciate emissions of a specific source/device/process.

4.4.4 Extrapolation

Extrapolation factors are usually used to estimate emissions for those source categories for which a SIC code has been assigned and for which employment data (typically by SIC) at the local level are available. Generally, this includes SIC categories 20-39 for industrial categories. In most cases, a large fraction of the VOC emissions within SIC's 20-39 will be covered by facility source procedures, so the extrapolation approach can be considered a secondary procedure to cover emissions from sources that are below the facility source size cutoff.

There are a number of source/device/processes for which emissions may be calculated as both a facility source and an area source. This is a result of U.S. EPA data reporting requirements, which specify that sources over a certain size should have their emissions calculated using the facility source approach (i.e., treated as facility sources instead of being lumped into an area source category). Facility source information is usually maintained in a database separate from area sources. For example, a database may contain information only on a few large industrial dry cleaners with VOC emissions over a certain emission threshold; emissions from all other industrial dry cleaners are calculated using the area source approach.

While emissions for the area sources can be calculated using regional activity data and subtracting the facility source portion (as discussed in Section 4.3.1), another method is to use information obtained by extrapolating data contained in the facility source database. In the extrapolation method, employment or some other appropriate activity functions as a surrogate activity level indicator for the activity of interest.

4.4.5 Engineering Calculations

Engineering calculations involve the use of chemistry and physics, principles, plus information on the design of the unit of operation or equipment, or emission information from similar processes to calculate emissions. Examples of engineering calculations include:

• The U.S. EPA Tanks model Version 2 (TANKS2);
• The U.S. EPA Surface Impoundment Model (SIMS);
• Refinery and chemical process unit fugitive emissions using organic vapor analysis (OVA) data; and
• Unique coating operations such as automobile assembly line coating.

Information on any engineering calculations that may be applicable to device/processes in other states should be made available through the Technical Steering Committee. Some engineering calculations have been computerized. These computerized tools are listed in Section 7.3. Table 4-8 describes the work breakdown for engineering calculations.

4.4.6 Direct Measurement

Emissions data are also available through direct measurement using continuous emissions monitors, usually located in the exhaust downstream of a combustion device. Information available from these devices can only be considered reliable if they are subject to a quality control/quality assurance (QA/QC) program that includes appropriate calibration.

Before accepting the data generated by this method, the state needs to verify that the monitors meet accepted criteria and that a QA/QC program is in place and functioning to ensure the quality of the data.

4.4.7 Process Simulation Software

A new technique for estimating emissions employs complex software to do real-time simulation of a process or group of processes. These systems were developed, not for emissions data development purposes, but to optimize complex industrial processes; emissions data are just one of the outputs that the system can generate. These systems use a variety of measurements, including emissions data, to develop relationships among the measured variables. Once initialized, the system runs in real-time to constantly adjust the process to reach a desired operational goal. Where these systems have been used to estimate emissions data, they have been able to generate data as accurate as any of the other available methods.

Before accepting the data using this method, the state needs to verify that the system has been shown to replicate real data and that the system was initialized over the entire operational range of the processes for which the system was installed.

4.4.8 On-Road Motor Vehicle Methods

On-road motor vehicle emissions are a specific type of area source category. Emissions can be calculated for each road or "link" and then summed to provide area-wide estimates, typically on a county-wide basis, assuming the information required to perform these calculations is available at that level of detail.

Motor vehicles are an important source of air toxic emissions. Certain uses of the emissions data derived from motor vehicles, such as risk assessments or deposition modeling, require spatially and temporally resolved emission estimates. Calculating emissions on a link-by-link basis will provide the most accurate emissions estimates and spatial distribution of these estimates across a given region.

Since the regional inventory will be compiled each year (at least initially), the following procedures recommended to compute motor vehicle emissions do not focus on techniques that lend themselves to temporally resolving these estimates (i.e., generating hour-by-hour estimates). These techniques are more resource intensive and are not warranted at this time.

The regional database will be capable of storing motor vehicle emission estimates on either a link-by-link basis, or on an area-wide basis. The following EET can be used to estimate emissions on either basis, although the data needs to support link-by-link emission estimation are substantially greater than computing emissions on a county basis. In compiling the regional inventory, it is recommended that emission estimates be computed on a link-by-link basis. At least four of the eight Great Lakes States (Illinois, Indiana, Michigan, and Wisconsin) participated in the Lake Michigan Ozone Study (LMOS). The VOC, NO_x, and CO emissions were computed on a link-by-link basis for that study. These criteria pollutant emissions are an important component of the EET for air toxics. Therefore, although the data requirements are substantial for link-by-link emission estimation, most of the required information has already been compiled, thus reducing the resources required.

The most accurate techniques to calculate motor vehicle emissions treat "trip-end" (i.e., start-up and evaporative park) emission separate from running (i.e., exhaust) emissions.

Calculating trip-end emissions requires information about the distribution of trips across a region, broken down by hot and cold start, and park duration. This type of detailed information is not typically available in a form suitable for emission estimation, particularly in the Great Lakes Region. The U.S. EPA, recognizing the difficulty and resources associated with developing appropriate trip-end data, have developed an EET for motor vehicles that incorporates trip-end emissions into an emission factor that is based primarily on the vehicle miles traveled (VMT) by different classes of motor vehicles acrossthe region. While we believe that trip-end emissions should be estimated separate from running emissions, the difficulty in generating such estimates at this time supports the recommendation that U.S. EPA procedures be followed.

Because of the complex nature of on-road vehicle emissions, they are not estimated using direct emission factors. Estimates of specific VOC or particulate compounds are derived from estimates of total hydrocarbons (THC) [THC is equivalent to TOG] or total particulate matter (TPM) [TPM is equivalent to PM]. The THC and TPM estimates are dependent upon local conditions, and must be calculated on a region-specific basis. Once THC and TPM estimates are generated, specific toxic compound emissions are estimated using source-specific speciation profiles. This process can be expressed in two equations. Emissions of toxic VOC pollutants are estimated using the general equation:

Emissions_species A = VMT * EF THC * wt%_{species A} / 100

where:

VM T= Activity data, VMT;

EF_THC=Total hydrocarbon emission factor; and

wt%_{species A}=Weight percent of species A obtained from appropriate species profile.

Emissions of particulate toxic pollutants are estimated from a similar equation:

Emissions_{species A} = VMT * EF_TPM * wt%_{species A} / 100

where:

EMT = Activity data, vehicle miles travelled;

EF_{TPM=Total particulate emission factor; and}

wt%_{species A}=Weight percent of species A obtained from appropriate species profile.

As stated above, the estimation of on-road motor vehicle emissions is complicated by the dependence of the emission factors for VOC and PM on local conditions. These local conditions include vehicle speed, ambient temperature, the use of inspection/maintenance (I/M) programs, and fuel characteristics.

Emission Factor Development - Emissions from on-road motor vehicles vary by the type of vehicle and the type of fuel used (gasoline or diesel). The THC and TPM for individual vehicle classes are calculated using the U.S. EPA's MOBILE5a and PART5 models. These models generate emission factors for each of eight vehicle classes:

• Light-duty gasoline vehicles (LDGV);
• Light-duty gasoline trucks, under 6,000 pounds gross vehicle weight (GVW) (LDGT1);
• Light-duty gasoline trucks, 6,000-8,500 GVW (LDGT2);
• Heavy-duty gasoline trucks (HDGV);
• Motorcycles (MC);
• Light-duty diesel vehicles (LDDV);
• Light-duty diesel trucks (LDDT); and
• Heavy-duty diesel vehicles (HDDV).

Of the two models, MOBILE5a requires the most local data. To run the model, the required inputs will include:

• Calendar year being modeled;
• Average vehicle speed;
• Minimum and maximum daily temperature;
• Fuel Reid vapor pressure;
• Vehicle operating model VMT distribution;
• Local registration distributions; and
• Specifications of any I/M or anti-tampering programs.

The U.S. EPA is finalizing the PART5 model for estimating particulate emissions from on-road motor vehicles. This model is similar to MOBILE5, although fewer local data are required. If the PART5 model is not available, then a national set of TPM emission factors are available from U.S. EPA (U.S. EPA, 1993a). Separate factors are available for diesel and gasoline particulate. The diesel factors, shown in Table 4-9, vary by calendar year due to the introduction of lower particulate emission standards.

Gasoline particulate emission factors are expressed as a fraction of THC emissions. Data appear to show a consistent correlation between gasoline particulate emissions and THC emissions. Gasoline particulate emissions are roughly 1.1% of THC emissions (U.S. EPA, 1993a).

Motor Vehicle Activity Data - The emission factors obtained from MOBILE5a and PART5 models are vehicle type-specific and dependent upon local conditions such as vehicle speed. To use these factors, the activity data must be developed in a manner consistent with these factors. Activity data for on-road motor vehicle emissions are expressed as VMT. Link-by-link annual average daily VMT should be obtained from the Highway Performance Monitoring System (HPMS) data or from transportation models. These data are disaggregated by area type (large urban, small urban, and rural) and by road type (also referred to as functional class). To properly model vehicle emissions, average vehicle speeds specific to each link should be used. These average speeds can be obtained from State Transportation Departments or local Metropolitan Planning Organizations (MPOs).

Once the data have been disaggregated on a link-by-link basis, they must be further disaggregated into VMT by vehicle class. Local data on vehicle class VMT fractions should be used if they are available. If not, then use national default values from MOBILE5.

Note that VMT data from HPMS and most other sources represent annual average values, which are adequate for the purposes of compiling the regional inventory of annual emission estimates. These annual average vehicle activity data must be adjusted to reflect the modeling conditions using seasonal or monthly adjustment factors if more temporally resolved emission estimates are desired. Depending upon the region, summertime VMT is significantly higher than the annual average. Further, if specific days of a week are being modelled, then additional adjustments must be made to reflect each day. In the summer, for example, both urban and rural VMT can vary considerably by the day of the week. Urban VMT is highest on weekdays, generally peaking on Fridays, with significantly reduced activity on weekends. In rural areas, particularly vacation centers, the opposite trend can be true, with significantly increased VMT on weekends versus weekdays. State transportation departments and local MPO's should be consulted for these adjustment factors. The techniques described above provide daily, seasonally adjusted emission estimates. If hourly emission estimates are needed, the transportation models could again be used to help provide this resolution. Many transportation models compute either hourly VMT estimates, or divide the day into a smaller number of discrete time periods, such as morning peak commute, afternoon peak commute, and off-peak. As the level of temporal disaggregation is increased, the computational and labor requirements (for QA/QC, for example) increase as well.

In dealing with a multi-state domain such as the Great Lakes region, the domain will include some areas where geocoded transportation models are available, andother areas where only county-wide VMT are available. It will be necessary to combine multiple data sources to produce a comprehensive, spatially resolved data set of on-road motor vehicle activity.

The computational requirements needed to compute emissions estimates on a link-by-link basis are considerably greater than when preparing county-wide emission estimates. A separate model such as the Motor Vehicle Emissions Model (MoVEM) (Radian Corporation, 1991) included in the Geocoded Emissions Modeling And Projections (GEMAP) system, was developed to assist the states participating in the LMOS in preparing gridded, hourly emission estimates for input to the Urban Airshed Model). MoVEM is capable of being extracted from GEMAP and run independent of the latter software. RAPIDS will include a separate source-specific model to estimate on-road air toxic motor vehicle emission estimates on a link-by-link basis.

Motor Vehicle Speciation Profiles - The final step in estimating toxic emissions from on-road motor vehicles is to speciate the THC and TPM emissions into individual compounds. For this activity, speciation profiles from the U.S. EPA's SPECIATE database should be used. The appropriate speciation profiles for each vehicle class are summarized in Table 4-10.

Table 4-11 summarizes the work breakdown structure for on-road motor vehicle methods.

These profile assignments use the latest speciation profiles developed from the Auto/Oil Study data. They include separate profiles for gasoline exhaust and gasoline evaporative emissions. To take advantage of these profiles, separate estimates of exhaust and evaporative emissions should be generated.