Skip to content U.S. Department of Transportation/Federal Highway AdministrationU.S. Department of Transportation/Federal Highway Administration

Office of Planning, Environment, & Realty (HEP)
PlanningEnvironmentReal Estate

HEP Events Guidance Publications Awards Contacts

Recent Examinations of Mobile Source Air Toxics

A Methodology for Evaluating Mobile Source Air Toxic Emissions Among Transportation Project Alternatives


Michael Claggett, Ph.D.
Air Quality Modeling Specialist, Federal Highway Administration Resource Center, 12300 West Dakota Avenue, Suite 340, Lakewood, Colorado 80228

Terry L. Miller, Ph.D., P.E.
Associate Professor, Department of Civil and Environmental Engineering, 73B Perkins Hall, The University of Tennessee, Knoxville, Tennessee 37996-2010


With the final update to its on-road mobile source emission factor model, MOBILE6.2, the U.S. Environmental Protection Agency (EPA) added capabilities of predicting emission factors for a select number of mobile source air toxics (MSAT), commonly referred to as the six priority MSATs. These are acetaldehyde, acrolein, benzene, 1,3-butadiene, diesel particulate matter, and formaldehyde. This presentation describes a methodology for computing and evaluating emissions of MSATs among a group of transportation project alternatives. The suggested scale of analysis is the affected transportation network, defined as those links where the annual average daily traffic is expected to change by ±5% or more as a result of the project. This analysis scale is considered reasonably representative of the regional scale emission factors predicted by MOBILE6.2. To gauge how emissions could change over an affected transportation network, provided are calculation ranges of MSAT emission factors produced by the model due to changes in a variety of input parameters. These include calendar year, ambient temperature, fuel Reid vapor pressure, and vehicle speed. Finally, a technique is presented for assessing MSAT emissions from the affected transportation network considering their relative toxicities. The technique allows a way to gauge the importance of increases and decreases in individual MSAT species amid proposed transportation alternatives and/or mitigation measures.


This paper provides the results of an analysis of air toxic emissions due to mobile sources for a hypothetical transportation project designed to mitigate traffic congestion. The example project involves the expansion of an existing urban freeway, plus upgraded arterial/collectors and freeway ramps to improve vehicular access. It is assumed that the freeway corridor extends 10 miles and that arterials cross the freeway every 2 miles with freeway/arterial access provided by freeway ramps. A No-Action Alternate was evaluated for the calendar year 2005 (present); the No-Action and two Build alternates were evaluated for calendar years 2010 (estimated time of completion) and 2030 (design year). The following notation/description is used in referring to the alternatives:

The underlying purpose of this effort is to provide a practical example of how a mobile source air toxics analysis may be applied to a planned project. This exercise offers additional insight into the technical challenges involved, including the formulation of analysis techniques; the types and sources of data required to complete such an analysis; the assumptions that may need to be made; the data forecasting routines and issues involved; and the comparative results likely to be obtained.



The basic procedure for conducting an emissions analysis or emissions inventory for on-road mobile sources is to calculate emission factors using the Environmental Protection Agency's (EPA) MOBILE model (EPA, 2003), then multiply by the vehicle-miles of travel (VMT) for each affected roadway link. The EPA's current version of the model, MOBILE6.2 (dated November 2003), is capable of predicting composite emission factors of the six priority mobile source air toxics (acetaldehyde, acrolein, benzene, butadiene, diesel particulate matter, and formaldehyde) in units of g/VMT. Most MOBILE6.2 emission factors are sensitive to changes in vehicle activity parameters so that the appropriate emission factors for a link are matched to the corresponding VMT/day. The sum product (g/VMT x VMT/day) for all affected links is obtained to provide emissions by pollutant on a ton per day or ton per year basis.

The mobile source emission factors predicted by the MOBILE6.2 model are applicable to a regional scale not an individual project corridor. Consequently, an emissions analysis for a project should include links beyond the project corridor and evaluated with respect to its effect on the transportation system. The affected transportation network can be defined as those links where the annual average daily (AADT) traffic is expected to change by more than ±5% as a result of a project.

Key Assumptions

The core assumption made in developing the traffic data for the emissions analysis is that the existing freeway and crossing arterials are operating at capacity (e.g., the volume-to-capacity ratio, V/C = 1) during the peak hour. Lanes are added to relieve the traffic congestion anticipated in future years. A growth rate of 1.5% per year in hourly traffic volumes on the freeway and crossing arterials was assumed for the No-Action Alternates based on Bureau of Transportation Statistics data (BTS, 2003) for the most recent 5-year record available (1998 through 2002). A higher growth rate (i.e., 1.75% per year) in hourly volumes was assumed for the upgraded projects to account for redirected trips from the surrounding area that may be diverted to a new, more efficient facility. The maximum hour-by-hour V-to-C ratios allowed on the facilities were 1.25 for the freeway and 1.15 for the crossing arterials. These are the major assumptions used to establish traffic volumes and speeds for the hypothetical upgrading projects.

In practice, a systems-level analysis would be required to adequately account for the redistribution of traffic on the upgraded project and on other parts of the affected transportation network as previously recommended. Or for projects located in relatively undeveloped areas, there is the potential for changes in surrounding land use and associated implications with respect to affected growth rates in traffic volumes. An actual systems-level analysis would need to account for this as well.

Traffic Data

Traffic activity data were developed based on methodology formulated by the Texas Transportation Institute (TTI) as provided in the National Highway Institute (NHI) course "Estimating Regional Mobile Source Emissions" and national data built into the MOBILE6.2 model. The capacity of the urban freeway is assumed to be 2100 vehicles per hour per lane (vphpl) (NHI, 2003; TRB, 2000) and the capacity of the urban crossing arterials is assumed to be 673 vphpl (NHI, 2003). Traffic volumes are assumed to vary hourly according to EPA's (2003) VMT fraction by hour of the day. For the 2005 existing condition, the roadways (i.e., freeway and crossing arterials) are assumed to be operating at capacity during the peak-hour traffic condition of 4 to 5 pm. Traffic volumes for the remaining hours are distributed based on the assumed capacity multiplied by a ratio of the VMT fraction for each hour divided by the VMT fraction for the peak hour. Total hourly volumes were determined considering the number of lanes associated with the existing condition, i.e., 6-lane freeway with 4-lane crossing arterials. A 50/50 directional split was employed. No distinction for weekend travel was made.

Traffic volumes for future years were determined by applying the assumed annual growth rate of 1.5% per year for the No-Action Alternate and 1.75% per year for the Build Alternates, limited to 1.25 x V/C for the freeway and 1.15 x V/C for the crossing arterials during any one hour. Capacity-limited volumes were only applicable for the 2030 No-Action Alternate. The resulting hourly traffic volumes are provided in Figure 1.

One reason for computing hourly traffic volumes is to determine hourly travel speeds, which vary according to the V-to-C ratio. The TTI method (NHI, 2003) for predicting congested speeds was applied. The basis for the methodology is calculating a congested speed (in mph) accounting for the effects of delay (min/mi) on the free-flow speed (in mph):

Congested Speed = 60 / (60/Freeflow Speed + Delay)

Default free-flow speeds are provided as a function of area type and roadway functional classification defined in the FHWA Highway Performance Monitoring System (HPMS). The default free-flow speeds for urban freeways and urban other principal arterials are 65 mph and 40 mph, respectively. The formula for calculating delay is:

Delay = Minimum[A x e<sup>B(V/C), M</sup>]

where A and B are volume/delay equation coefficients and M is the maximum minutes of delay per mile. Default values are provided: A = 0.015, B = 3.5, and M = 5 for high-capacity facilities; A = 0.05, B = 3, and M = 10 for low-capacity facilities. Default capacities are also provided as a function of area type and roadway functional classification: C = 2100 vphpl for urban freeways and C = 673 vphpl for urban other principal arterials. Locale-specific parameters should be derived and used in calculating congested speeds for most applications.

Figure 1. Hourly Traffic Volumes and Congested Speeds.

Hourly Traffic Volumes for the 6-Lane No-Action Alternative
click for data to all 6 images in Figure 1

Hourly Traffic Volumes for the 6- to 8-Lane Build Alternative
click for data to all 6 images in Figure 1

Hourly Traffic Volumes for the 6- to 10-Lane Build Alternative
click for data to all 6 images in Figure 1

Hourly Congested Speeds for the 6-Lane No-Action Alternative
click for data to all 6 images in Figure 1

Hourly Congested Speeds for the 6- to 8-Lane Build Alternative
click for data to all 6 images in Figure 1

Hourly Congested Speeds for the 6- to 10-Lane Build Alternative
click for data to all 6 images in Figure 1

The resulting travel speeds are given in Figure 1 as previously referenced. An average hourly congested speed for the day was also computed to determine if it may be used as a surrogate for an hour-by-hour variation in speeds. The average hourly congested speeds illustrated in the figure are applicable to all hours of the day, but only a portion of each series is presented so that the hourly congested speeds can be more clearly shown. The hourly congested speeds predicted encompass the average speeds of the test cycles used in developing the speed correction factors in MOBILE6.2, i.e.:

The daily VMT is the product of the Annual Average Daily Traffic (AADT) and the facility length. The hourly volumes by facility type were summed to obtain the AADT as provided in Table 1 by alternate. The facility lengths assumed are 10 miles for the freeway and 6 miles for the crossing arterials (i.e., 6 arterials of 1 mile in length each). The resulting daily VMT for each alternate are also presented in Table 1.

Identical hourly traffic volumes, AADT, and daily VMT are realized for the 6- to 8-Lane and 6- to 10-Lane Build Alternates. Even so, there are differences in the capacities and predicted congested speeds for the build alternates that may affect the respective MSAT emission totals. In contrasting the No-Action and Build Alternates, differences in hourly traffic volumes, AADT, and daily VMT are observed due to the AADT growth rates and V-to-C ratio limits implemented.

MOBILE6.2 Inputs

The MOBILE6.2 model was run using EPA's national default data built into the program with the following exceptions. Parameters for which there are no default values include calendar year; minimum and maximum temperature; gasoline fuel Reid vapor pressure (RVP); average diesel fuel sulfur level and maximum particle size cutoff (for diesel particulate matter); and specifications of the gasoline fuel used (for acetaldehyde, acrolein, benzene, 1,3-butadiene, and formaldehyde). Parameters for which national defaults were not used include month of evaluation and average speeds. Emission reductions that may be realized from a local inspection/maintenance program were not taken into account.

The calendar years evaluated include 2005 as the baseline year; 2010 as the estimated time of completion; and 2030 as the design year. When conducting annual emissions inventories, EPA recommends that monthly emission factors be developed via mathematical interpolation between January and July and summing the monthly emissions results. To simplify this analysis, the parameters that would vary by month are represented by a single value as the basis for the annual emissions inventory. An evaluation of the variability of MOBILE6.2 emission factors is provided to gauge how changes in certain assumptions would affect emission factors representative of freeway and arterial operation.

The MOBILE6.2 model was run assuming no temperature variation over the day simulated (i.e., minimum temperature = maximum temperature) using a temperature of 55 ºF to represent an annual average. The median of the annual average daily minimum and maximum temperatures measured in the U.S. are 43.3 ºF and 63.6 ºF, respectively; 55 ºF is about midway

Table 1. Travel Characteristics of Each Alternate.

Annual Average Daily Traffic (vpd)
Alternate Year
2005 2010 2030
Freeway Arterials Total Freeway Arterials Total Freeway Arterials Total
6-Lane No-Action 162162 34646 196808 174695 37324 212018 229669 47963 277632
6- to 8-Lane Build       176857 37786 214642 250213 53458 303671
6- to 10-Lane Build       176857 37786 214642 250213 53458 303671
Daily Vehicle-Miles of Travel (VMT per day)
Alternate Year
2005 2010 2030
Freeway Arterials Total Freeway Arterials Total Freeway Arterials Total
6-Lane No-Action 1621622 207876 1829498 1746947 223942 1970889 2296691 287779 2584470
6- to 8-Lane Build       1768567 226713 1995281 2502131 320749 2822880
6- to 10-Lane Build       1768567 226713 1995281 2502131 320749 2822880

between these values. The median of the normal daily minimum temperatures measured in the U.S. during the coldest month of the year (January) is 23.5 ºF and the median of the normal daily maximum temperatures measured in the U.S. during the warmest month of the year (July) is 86.1 ºF. The fuel RVP would change over the course of a year from the switching of winter fuel blends to summer fuel blends and back again. The range of fuel RVP in some locales can be expected to encompass the volatility of class AA (7.8 psi) through class E (15.0 psi) fuels prescribed by the American Society of Testing Materials (ASTM). A fuel RVP of 12.5 psi (Class C/D) was assumed for this analysis. The evaluation month selected was July.

Emission factors of diesel particulate matter include the organic carbon, elemental carbon, and sulfate portions of diesel exhausts for a maximum particle size cutoff of 10 µm. The diesel fuel sulfur levels used are consistent with the 49-state average values reflecting more stringent federal controls (i.e., 11 ppm for 2010 and 2030). For the baseline year of 2005, an average diesel fuel sulfur level of 350 ppm was assumed. Emission factors for the hydrocarbon MSATs were based on the 2007/2020 30 ppm fuel specifications for the northeastern states during summer and no reformulated fuel program (RFP).

The emissions analysis was conducted by accounting for the vehicle emission types specific to the operation of the facility, e.g., exhaust running and evaporative running loss emissions for vehicles operating on the freeway and crossing arterials. The national defaults for start and soak emissions built into the MOBILE6.2 model are not applicable to a project-level analysis as most of the starts and ends of vehicle trips would not occur on the upgraded project or on the affected transportation network. Start and soak emissions need to be accounted for if a project would significantly increase the number of trips above the No-Action Alternate, not just a redistribution of existing trips.

Updated: 07/06/2011
HEP Home Planning Environment Real Estate
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000