Introduction

Final Report
Prepared for

U.S. Department of Transportation
Federal Highway Administration
Office of Natural Environment
Washington, DC 20590

Prepared by:

Volpe National Transportation Systems Center
U.S. Dept. of Transportation
Research and Innovative Technology Administration
55 Broadway
Cambridge, MA 02142-1093

This document discusses the sensitivity of various input parameter effects on emission rates using the US Environmental Protection Agency's (EPA's) MOVES2010a model at the regional level. Pollutants included in the study are carbon monoxide (CO), Oxides of Nitrogen (NOX), Particulate Matter of less than 2.5 micrometers (PM2.5), and Volatile Organic Compounds (VOCs). Similar trends for PM10 as reported for PM2.5 and Nitrogen Dioxide (NO2) as NOX exist and inferences to these pollutants may also be made. Results are presented using the predicted emission rates (grams/mile) for running exhaust and starts across multiple MOVES source types.

December 10, 2012
Report Number: DOT-VNTSC-FHWA-12-05

Full Report (PDF, 3.6 MB)

Executive Summary

This document discusses the sensitivity of various input parameter effects on emission rates using the US Environmental Protection Agency's (EPA's) MOVES2010a1 model (20100830 database) at the regional level. Pollutants included in the study are carbon monoxide (CO), Oxides of Nitrogen (NOX), Particulate Matter of less than 2.5 micrometers (PM2.5), and Volatile Organic Compounds (VOCs). Similar trends for PM10 as reported for PM2.5 and Nitrogen Dioxide (NO2) as NOX exist, and inferences to these pollutants may also be made. Results are presented using the predicted emission rates (grams/mile) for running exhaust and starts across multiple Motor Vehicle Emission Simulator (MOVES) source types.

The input parameters varied in this analysis are: Temperature, Humidity, Ramp Fraction, Age Distribution, Analysis Year, and Average Speed Distribution. The input parameters of Road Type Distribution, Source Type Population, Age Distribution, Fuel, and Inspection and Maintenance (I/M) Programs were held constant utilizing the national default values from the MOVES 20100830 default database for the 2010 Analysis Year. MOVES is a complex model with many input parameters that can influence the emission rates across multiple vehicle types. The overall modeling process may include many variations and is not covered by this report. A separate project level analysis will delve more into the overall modeling process.

The results of the model sensitivity are presented for various vehicle types utilizing particular fuel types to provide an understanding of the input sensitivity independent of fleet mixture. The emission rate values are included in the results tables located in Appendices allowing the user to review the magnitude of the emissions rates across vehicle types. These data are specific for this sensitivity analysis and are not meant as absolute values for use in regional emissions analyses.

The methodology of the analysis used a local sensitivity analysis approach where a single input parameter was varied while all the other input parameters were held constant. The output emissions rates were analyzed across all MOVES vehicle types. To allow a comparison of these emission rates, a 'Baseline Case' was established. The Baseline Case used the default data from a National Scale MOVES run allowing national defaults for road type distribution, age distribution, average speed distribution, fuel, ramp fraction, and I/M programs. In order to run MOVES in a time efficient manner, a surrogate model approach was utilized to represent a county level analysis while executing MOVES for a single hour of the day. The surrogate approach utilizes a less computationally expensive method of running MOVES to obtain the overall sensitivities. A single hour was sufficient to establish the trends associated with the various model sensitivities as input parameters were varied.

While described in detail within the report, the basic findings for each evaluated parameter are presented:

• Temperature is a very sensitive parameter across all pollutants and vehicle types. The results from this analysis showed similar trends to the temperature and humidity sensitivity analysis conducted by EPA.
• Analysis Year is a very sensitive parameter especially between the years 2010 and 2020 where emission rates are seen to decrease most significantly. Emission rates further decline until the year 2040 and remain relatively unchanged thereafter. Given the analysis year requirements, prescribed for regional conformity determinations, users may not have a lot of flexibility in varying this input parameter.
• Age Distribution of the vehicle fleet is important. A proportional increase of 10 percent in the distribution of vehicles less than 10 years old caused a reduction in vehicle emission rates by approximately 16 percent for CO, 12 percent for NOX, and 11 percent for PM2.5. As expected, an older fleet with a 10 percent greater distribution of vehicles between 11 and 20 years old resulted in an increase in emission rates across all pollutants. This trend continued when increasing the proportion of the oldest set of vehicles between 21 and 30 years old as well. It is desirable for the users to obtain local vehicle age distribution data instead of relying on default information. This is especially true if the area's fleet consists of newer vehicles or if vehicle replacement programs are in effect.
• Ramp Fraction can be a sensitive input parameter dependent on vehicle and fuel type. A common observation for almost all vehicle types across all pollutants was that emission rates and ramp fraction change in a linear manner. As the ramp fraction increases, so do emissions rates. Diesel emissions of CO remained relatively flat showing a dependence on fuel type within the model. Alternatively, the emissions rate for PM2.5 showed an increase for diesel fueled vehicles with increased ramp fraction while gasoline emissions remaining somewhat constant. This parameter will be greatly controlled by the highway geometric design.
• Emission rates for NOX and CO were the most sensitive pollutants due to changes in humidity. In the case of CO, gasoline fueled vehicles showed increased emissions as humidity increased, while for NOX, diesel fueled vehicles were most affected. All other vehicle types remained relatively insensitive to changes in humidity.
• The emission rates associated with Average Speed Distributions representing Level-of-Service (LOS) B, C, and D generally varied by only a few percentage points across all pollutants and vehicle types. Results for CO varied for all vehicle types and should be examined individually by the reader in the full report. The emissions rates associated with LOS E showed a larger variation than LOS B, C or D, while emission rates associated with LOS F were significantly higher. The 'Baseline case' exhibited an emission rate between LOS E and F. Use of default values results in a LOS E+ speed to volume relationship, which in turn indicates a conservative bias for the MOVES default values. This is an indication that local data should be obtained and used when possible. The functional classification for arterials show a much greater change in emission rates for varying LOS than all other facility types.

The analyst should be aware of how all of these variables affect a regional analysis and the information of this report should inform in that regard. This provides an awareness of the importance of inputs during the design phase of projects and could result in a better analytical design in regards to air quality. Default data or assumptions should not be used if it is possible to obtain local data. This is especially true for vehicle age distribution and average speed distribution with related drive schedules. For example, defaulting to the MOVES average speed distribution would result in a LOS E+ being used during analysis. This heavy congestion may not exist or may not be the outcome of a final design and if used could result in higher emission rates than would occur if the actual speed distribution were used. Temperature and humidity are location specific. The analysis year will be defined by conformity guidelines. Omitting these two input parameters, the order of impact for including actual data would be:

• Average speed distribution for arterials
• Vehicle age distribution
• Ramp fraction
• Average speed distribution for interstates
• Average speed distribution for freeways.

Use of local data inputs is generally considered to provide the most accurate on-road mobile source emissions estimates.