The U.S. Environmental Protection Agency (EPA)'s Motor Vehicle Emission Simulator (MOVES) has replaced MOBILE6 as a regulatory emissions model for use in State Implementation Plan (SIP) development and transportation conformity analyses in all States, except California. With several methodological improvements, MOVES appears to be more data intensive as compared to MOBILE6. Therefore, preparing local input data for model runs in MOVES can be challenging, if not demanding. Although much of the MOVES data requirements may be converted from existing data used in MOBILE6, some data still need to be newly developed or further refined.
The modeling concepts and methodologies in MOVES are also significantly different from those in MOBILE6. For example, the functional design concepts in MOVES disaggregate emission sources primarily by source use type (e.g. passenger car, passenger truck, single-unit long-haul truck, etc.), each of which are further categorized into several source bins by several characteristics including model year, fuel type, engine technology, loaded weight, engine size, and regulatory class. As another example, the basis of vehicle activity for exhaust running emissions is source-hours operating (SHO) rather than vehicle-miles traveled (VMT) that have been used in the MOBILE model series. These modeling changes can be taken as a good opportunity for the transportation and air quality community to explore new data sources and to develop new methods for utilizing existing data sources to improve the emission inventory and analysis of on-road mobile sources.
In general, on-road emission modeling requires three types of data: 1) vehicle fleet data, 2) vehicle activity data, and 3) vehicle emission rate data. The current practices, trends, and future needs on how each of these data types is generated and used are briefly discussed below.
In terms of vehicle fleet data, it has been recognized that in the development of an emissions inventory, reliable data of vehicle fleet population such as vehicle class and age distributions are as important as accurate data concerning vehicle emission factors and vehicle activities. However, this is probably the area that has been given the least amount of attention by practitioners and researchers.
The current practice of deriving vehicle fleet data relies heavily on the use of vehicle registration databases. These databases are sufficient for creating base vehicle fleet distributions for an area, but the sole dependency on this type of data is problematic for several reasons. For example, a significant fraction of vehicle miles traveled (VMT), and thus emissions, for an area could be attributable to vehicles registered outside the area [Malcolm et al., 2003; Lutsey, 2009]. This is particularly true for tourist areas and areas with major transportation hubs (e.g., international airports and seaports). The bias in vehicle fleet distributions could have significant impacts on both SIP and transportation conformity analyses. Another example is the inability to provide specific vehicle fleet distributions for short time periods (e.g. hours of day, months of year).
Therefore, there is a need for tools and methods that will enable the collection and development of highly-resolved and area-specific vehicle fleet data.
When constructing emission inventories for use in SIP development and transportation conformity analyses, several vehicle activity data inputs are required. For instance, MOVES requires VMT, average speed distribution, number of trip starts, soak time distribution, among other inputs. These data inputs need to be generated for all vehicle types and, depending on a particular data input, may need to be characterized by road type, month, day, and hour. Acquiring all these data inputs are not easy as the availability of existing data is limited, especially for non-passenger vehicles such as heavy-duty trucks (HDTs).
Albeit a very small fraction in the total vehicle population, HDTs contribute disproportionately to the emissions inventory of on-road mobile sources. This is due to their high annual mileage and high emission rates. In addition, HDTs are also a significant source of idling emissions especially at truck stops and terminals as they often engage in long-duration idling activities (e.g., loading/unloading, heating/cooling the cabin during rest stops, etc.) at these locations [Miller et al., 2007; Frey et al., 2008]. Therefore, an accurate characterization of HDT activity is crucial to the construction of emissions inventory of on-road mobile sources.
In the current state of the practice, the Highway Performance Measurement System (HPMS) has been used as a primary source for VMT data for various road and vehicle types, including HDTs [U.S. Environmental Protection Agency, 2005]. Although HPMS contains roadway speed limit information, it does not provide measured traffic speed data. Therefore, the reported VMT cannot be characterized by speed bins. As vehicle emissions are sensitive to vehicle speed among other things, it is desirable to characterize VMT into multiple speed bins so that appropriate emission factors for each speed bin can be applied.
Alternatively, HDT activity can be estimated using travel demand models, especially those with a dedicated module for HDTs (e.g., [Southern California Association of Governments, 2008]). There has also been increasing interest in developing freight flow models (e.g., [Sarvareddy et al., 2005]), which can be used to derive truck trips and miles traveled. Nevertheless, these models are still in their early stages and have not been adopted widely. Also, the availability of measured truck traffic data, especially with regards to speed, that can be used for model validation is limited so that the accuracy of speed data from the models may be questionable.
Another method that has been used is to instrument a fleet of HDTs with GPS-based data loggers and log their travel activity over a period of time (e.g., [Battelle, 1999]). This method offers the most detailed and probably the most reliable information on HDT miles and speed. Also, it is able to capture the information about non-driving activities such as soak time and idling, which are not available in either the HPMS or travel demand models. However, this type of data collection requires significant resources; and thus, is usually performed for a small number of trucks and for a short period of time.
Thus, there is a need to explore alternative sources of HDT activity data that have not been used in the past, but have potential to be useful. Also, it is desirable to develop new and innovative methods for utilizing existing data sources to their full potential.
In terms of vehicle emissions, although the development of MOVES is based on a wide variety of measured energy and emissions datasets, there are still approximately 50% data "holes" [U.S. Environmental Protection Agency, 2005]. For instance, emissions data of HDTs of the recent model years were not available by the time MOVES was developed. These holes have been filled using hole filling methods. It is assumed that as more emissions data from vehicles are collected, the emissions rates in MOVES can be updated and thus the fidelity of the model should improve.
It should be noted that not all data holes are equally important. For instance, data holes for emissions sources that do not contribute much to an emissions inventory (e.g., diesel passenger car) will only make negligible differences to the inventory even though they may be significantly underestimated or overestimated. On the other hand, some data holes will have significant impacts on the resulting emissions inventory as well as the policy implications. One example is the energy/emission rates of HDTs with various loaded weights. It can be argued that not all VMT of HDTs carry the same amount of weight, and it is intuitive that the higher the loaded weight, the higher the amount of engine power required, and thus the higher emission rates.
Hence, there is a need to continue measuring vehicle emissions in order to fill in the emission data holes in MOVES. These emission measurement programs should be prioritized so that critical data holes are addressed first.
The goal of this research is to improve the estimation or measurement of the three data types necessary for on-road emission modeling, namely vehicle fleet data, vehicle activity data, and vehicle emission data. For each of these data types, this research targets specific areas or data elements that are considered gaps in the current state of the knowledge and practices by evaluating alternative data sources as well as developing new tools and methods for filling such gaps. Specifically, the objectives of this research are to:
This report is organized into five chapters as follows:
1. Chapter 1 is the introduction, which briefly describes the topics being addressed in this research.
2. Chapter 2 reviews the current state of the practice in estimating vehicle class and age distributions, VMT of HDTs, and idling hours of HDTs from selected agencies around the Nation.
3. Chapter 3 presents the method for deriving local vehicle fleet data through the use of license plate survey in conjunction with vehicle registration database and VIN decoder.
4. Chapter 4 consists of three parts. The first part describes the use of truck's ECU data to generate some HDT activity data inputs for MOVEs. The second part describes a large-scale truck telematics dataset and demonstrates methods for generating HDT activity data inputs for MOVES based on this dataset. In the third part, a methodology is presented that combines data from weigh-in-motion stations and vehicle detector stations on freeways to generate highly resolved HDT activity data.
5. Chapter 5 describes an emission testing program that measures emissions from two HDTs meeting the 2007 emission standard at different loaded vehicle weight.
The last chapter is followed by a list of references and a series of appendices that show detailed results from the various analyses in this research.
The report is written in a way that each chapter can be a standalone document. They start with background information on the topic being address, followed by a detailed presentation of data used, methodology, and results, and then end with concluding remarks specific to that chapter.