Project Proposal

Background

Air Quality Modeling

The sources of air pollution are broadly classified into stationary sources, mobile sources and naturally occurring sources.    Factories, power plants, dry cleaners and the like fall into the category of stationary sources; mobile sources include cars, buses, planes etc; and windblown dust and volcanic eruptions constitute naturally occurring sources. These sources emit a variety of pollutants which are classified by the EPA as 6 principal pollutants, also known as criteria pollutants. The EPA along with several national, state and local organizations monitors the level of these pollutants in the atmosphere. Carbon monoxide (CO), Nitrogen Oxides (NO_x), Lead (Pb), Ozone (smog), particulate matter (PM10) and sulphur dioxide (SO₂) are the six criteria pollutants addressed by the NAAQS (National Ambient Air Quality Standards).

            The OAQPS (Office of Air Quality Planning and Standards) is responsible for setting the NAAQS and it monitors for the criteria pollutants through various programs. It is also responsible for ensuring that these air quality standards are met (or "attained"). If the level of these criteria pollutants is higher than what is considered acceptable by the EPA, then the area in which the level is too high is called a nonattainment area. In order to work towards attainment, the OAQPS requires that each state containing nonattainment areas develop a written plan (SIP: State Implementation Plan) for cleaning the air in these areas. The Clean Air Act states that no transportation activity can be funded or supported by the Federal Government unless it conforms to the purpose of a state's air quality plan. Conformity is a Clean air act requirement intended to ensure that new transportation investments do not jeopardize air quality in nonattainment and maintenance areas. The nonattainment areas in Texas include Dallas-Fort Worth, Beaumont-Port Arthur and Houston-Galveston-Brazoria (Ozone nonattainment areas); and El Paso (nonattainment area in Ozone, CO and PM10).

Tailpipe emissions form about a third of the air pollution in the country. The Office of Transportation and Air Quality (formerly the Office of Mobile Sources) provides a wealth of information on mobile source emissions including details about emissions modeling software and databases. For further information refer Vehicle and Engine Emission Modeling Software. Over the years several models have been developed to measure/estimate air pollution from vehicles, engines and fuels. These include the MOBILE series, PART5, Non-Road Vehicle & Engine Emission Models and Fuels Models.

MOBILEx Software

MOBILE, the USEPA's highway vehicle emission factor model, is a FORTRAN program that provides average in-use fleet emission factors for three criteria pollutants [volatile organic compounds (VOC), a precursor of ground-level ozone; carbon monoxide (CO); and oxides of nitrogen (NOx)], for each of eight categories of vehicles (detailed below), for any calendar year between 1970 and 2020 and under various conditions affecting in-use emission levels (e.g., ambient temperatures, average traffic speeds, gasoline volatility) as specified by the model user. The MOBILE5a input screen (DOS environment) is shown below.

The vehicle categories for which MOBILE calculates average in-use emission factors are: 

LDGVs light-duty gas vehicles (passenger cars), up to 6000 lb gross vehicle weight (gvw) 

LDGT1s light-duty gas trucks (pick-ups, minivans, passenger vans, and sport-utility vehicles), up to 6000 lb gvw 

LDGT2s LDGTs of 6001-8500 lb gvw (heavier versions of LDGT1s; the categories are modeled separately because numerically different emission standards are established under the Clean Air Act for LDGT1s and LDGT2s) 

HDGVs heavy-duty gas vehicles, which are vehicles of 8501 lb and higher gvw that are equipped with heavy-duty gas engines

LDDVs light-duty diesel vehicles (passenger cars), up to 6000 lb gvw 

LDDTs light-duty diesel trucks, up to 8500 lb gvw (unlike gasoline powered LDTs, the same emission standards are applicable to all diesel LDTs up to 8500 lb gvw) 

HDDVs heavy-duty diesel vehicles, vehicles of 8501 lb and higher gvw equipped with heavy-duty diesel engines 

MCs motorcycles (all of which are gasoline powered; highway-certified motorcycles only are included in the model, off-road motorcycles such as "dirt bikes" are modeled as a nonroad mobile source)

The model was first developed as MOBILE1 in the late 1970s, and has been periodically updated to reflect the collection and analysis of additional emission factor testing results over the years, as well as changes in vehicle, engine, and emission control system technologies, changes in applicable regulations and emission standards and test procedures, and improved understanding of in-use emission levels and the factors that influence them. MOBILE6 was released earlier this year and is geared to include: facility-based emission factor estimates (different average emissions for different roadway types, even at similar average speeds), needed for transportation conformity determinations and more sophisticated application of results (e.g., photochemical air quality modeling, as versus simple inventory tabulation); "real-time" diurnal emission factors; updates on effects of oxygenated fuels on CO emissions, and effects of in-use fuel sulfur content on all emissions; separation of "start" and "running" emissions, to permit more precise temporal and spatial allocation of emissions; updates to many other areas on basis of new data. For more details on MOBILE6 refer http://www.epa.gov/otaq/m6.htm.

MOBILE6 separates the 'start' and 'running' emissions in order to permit more precise temporal and spatial allocation of emissions. Inputs to the model include trip length estimates, trip start and trip ends, diurnal soak time, engine start soak time distributions, VMT by hour of day, facility, speed etc. The model then estimates 'emission factors' expressed as grams of pollutant per vehicle mile traveled (g/mi). This can be combined with estimates of total VMT to develop highway vehicle emission inventories (tons/day). The emission type classifications used in MOBILE6 are as follows

running emissions: exhaust running emissions

start emissions: exhaust engine start emissions (trip start)

hot-soak emissions: evaporative hot soak emissions (trip end)

diurnal emissions: evaporative diurnal emissions (heat rise)

resting emissions: evaporative resting loss emissions

run loss emissions: evaporative running loss emissions

crankcase: evaporative crankcase emissions (blowby)

refueling emissions: evaporative refueling emissions (fuel displacement).

In this project we focus on the engine start soak time distributions as input to the MOBILE6 model.

Research - Soak Time Models

Soak Time is defined as the duration of time in which the vehicle's engine is not operating and which precedes a successful vehicle start (i.e. one that does not result in a stall). A vehicle trip start with a soak time that exceeds 12 hours is referred to as a COLD START, and engine starts with shorter soak times are called HOT STARTS. MOBILE6 takes as input the soak time distribution which is discretized into 70 soak time bins for convenience. Soak time duration of 1 to 30 minutes is split into bins of 1 minute intervals; those of 30-60 minutes is split into bins of 2 minute intervals and soak times of 60-720 minutes is split into bins of 30 minute intervals. This demarcation is based on the emission levels associated with the different soak time durations. We use as input the fraction of trips in each of 70 soak time bins by time-of-day and origin activity purpose.

The research team at the University of Texas at Austin [Nair, Bhat & Kelly (2000)] has developed a model that predicts the zone-specific soak-time distributions for each combination of time-of-day and activity purpose. Given the zone-to-zone production-attraction matrices by trip purpose, we can obtain the number of trip starts by the zone of origin, activity purpose prior to trip start, and time-of-day (by applying fixed factors obtained from travel surveys). This can then be used to compute the zone-specific soak-time distributions using the soak time distribution models. First a binary logit model is applied to estimate the fraction of vehicle starts that are first trip starts. Log-linear soak time models (specific to first and non-first trip starts) can then be applied to estimate the fraction of trip starts in each soak time bin. From these models, soak time duration is seen to depend upon the time-of-day of the trip start, activity purpose preceding the trip start, land use and demographic characteristics of the zone of trip start and trip characteristics. The time periods and origin activity purposes used in these models are as follows.

Binary Logit Model

The fraction of first trip starts (which is equal to the probability of a vehicle start being a first start) is given by

d _tiz = P_tiz(G=1) = 1/(1+e^-[g_n^+b'x_z^])

where t: time of day

           i: origin activity purpose

          z: origin zone

Log-linear soak time models

The (log) soak-distribution of trip starts in time t with origin activity purpose i in zone z for first starts and non-first starts may be written as follows

The means and variance of these distributions can be estimated from the parameter estimates obtained in the estimation stage.  

The coefficients in the above models are estimated and can be picked up from the paper "Modeling Soak-Time Distribution of Trips for Mobile Source Emissions Forecasting: Techniques and Applications" [Nair, Bhat & Kelly (2000)]. The objective in our effort is to obtain the fraction of soaks in each of 70 time-bins (as needed by MOBILE6) across both first- and non-first starts for each zone, and for each activity purpose and time-of-day combination. Let k be an index for time-bin (k=1,2,…70), and let time-bin k be bounded by the continuous soak-time value of m_k-1 to the left and m_k to the right. Then the fraction of soaks in time-bin k for first starts may be written as:

The corresponding expression for non-first starts is

Finally, the fraction of soaks in time-bin k across first and non-first starts for zone z, time-of-day t and activity purpose i may be computed as follows

back to top

Objective

To implement the soak time distribution models in a GIS framework with the following features                                          - user interactive input                                                                                                                                                                                 - output format suitable for MOBILE6                                                                                                                                                        using TransCAD and GISDK (GIS Developer's KIT)

TransCAD

TransCAD is a GIS software, not unlike ArcView or ArcInfo but with the focus on solving transportation related problems. Accordingly it is replete with procedures that can solve transportation problems and apply transportation models all in a single integrated platform. It can be used for all modes of transportation, at any scale or level of detail. TransCAD is a state-of-the-art GIS that you can use to create and customize maps, build and maintain geographic data sets, and perform many different types of spatial analysis. It includes sophisticated GIS features such as polygon overlay, buffering, and geocoding, and has an open system architecture that supports data sharing on local- and wide-area networks. TransCAD extends the traditional GIS data model to include transportation data objects such as transportation networks, matrices, routes and route systems and linear-referenced data. These extensions make TransCAD the best tool for working with transportation data.

TransCAD also includes application development tools and user services. The Geographic Information Systems Developer's Kit (GISDK) gives the user the tools needed to create a wide variety of products for delivering mapping and geographic analysis capabilities. Over 700 functions can be called from Caliper Script, a complete macro-programming language for designing menus and dialog boxes (including toolbars and toolboxes), and for writing macros. GISDK can be used to create add-ins that extend the standard interface to provide new capabilities, custom interfaces that focus the user on the capabilities needed for a particular purpose and embedded desktop applications.

Study Area

For this project we shall focus on the Dallas-Fort Worth area at the TAP (Traffic Analysis Process) zonal level. Each TAP zone is made of several disaggregate TSZs (traffic survey zones). Zonal land use and demographic characteristics were obtained from the North Central Texas Council of Governments (NCTCOG) at the TSZ level and were aggregated to the TAP zonal level. Shapefiles of the Dallas-Fort Worth area were also obtained from the NCTCOG. The following datafiles and shapefiles form a part of this project.

Dallas-Fort Worth: TAP zones (919)

back to top

Procedure

The implementation of the soak-time distribution models in a GIS framework involved the following steps.

Apply the soak-time models in Excel. Use the coefficient matrix for the soak-time models together with the land use and demographic data to apply the models in Excel. This yields the soak-time distribution (fraction of trips in each soak-time bin) by origin activity purpose for each of the six time periods and for each tap zone in D-FW.

Create dataviews from the Excel results. The soak time distribution for each time period is imported into TransCAD as a dbf file and stored as a TransCAD dataview. This is done for each of the 919 TAP zones. For example; the soak-time distribution for the Morning period for TAP zone 4 is given here.

Nair, Bhat & Kelly (2000). Modeling Soak-Time Distribution of Trips for Mobile Source Emissions Forecasting: Techniques and Applications.

 

http://www.epa.gov/oar/oaqps/cleanair.html Office of Air Quality Planning and Standards.

 

http://www.epa.gov/otaq/models.htm Office of Transportation and Air Quality - Vehicle & Engine Emission Modeling Software.

 

http://www.caliper.com Caliper Corporation.

 

TransCAD User's Guide

 

http://www.dfwinfo.com/index.asp North Central Texas Council of Governments.