Project Summary Report 1838-S – 1 –

The University of Texas at Austin

C e n t e r f o rTransportation Research

P

RO

JE

CT

S

UM

MA

RY

R

EP

OR

TCENTER FOR TRANSPORTATION RESEARCH

THE UNIVERSITY OF TEXAS AT AUSTIN

Research Project Summary Report 1838-STransportation Control Measure Effectiveness in Ozone Nonattainment Areas

Authors:Chandra R. Bhat, Huimin Zhao, Yasasvi Popuri, Monique Stinson,

and Sara Poindexter

August 2001

Transportation Control Measure Effectiveness InOzone Nonattainment Areas

Transportation planners,traditionally used to focusing onregional travel forecasting, now havethe added responsibility of providingtraffic data to state air qualityagencies for use in mobile sourceemissions analysis. Coordinatedefforts among land use planners,travel demand planners, and airquality planners are needed to ensurethe provision of safe and efficienttransportation systems while alsoaddressing environmental concerns.This is particularly so becausemobile source emissions constitute amajor fraction of total atmosphericemissions. Consequently, manymetropolitan areas and states aredepending on transportation controlmeasures (TCMs) to reduce mobilesource emissions as part of an overallstrategy to reduce atmosphericemissions. Within this context, it isimportant to develop a procedure todetermine which TCMs (or combina-tion of TCMs) have the mostbeneficial impact in terms ofmobility, emissions, and cost. It isnecessary that such a procedure bemethodologically sound yet applica-tion friendly, and be capable ofanalyzing the effects of demographicchanges and transport policy actions.The current research contributestoward the development of such anintegrated transportation air qualityprocedure. Specifically, the researchhas the following objectives: (a)Refine regional travel models tomake them more sensitive to socio-demographic changes and TCMs, (b)Develop models for improvedsupplementary traffic input dataneeded by the MOBILE emissionfactor model, and (c) Integrate allrelevant models within a GeographicInformation System (GIS) architec-

ture. The next section discusses theresearch conducted in the project toachieve each of the objectivesidentified above.

What We Did...

Refine Structure and Specificationof Travel Models

Several areas of improvement intravel models were undertaken aspart of this project. These arediscussed below.

How many trips are produced?Two improvements were made

to traditional trip productionmodeling procedures. First, anordered-response discrete choicemodel that accommodates the ordinalbut discrete nature of trips wasestimated at the household-level.Using such a model structure istheoretically more appropriate thanusing cross-classification or linearregression approaches, since theselatter approaches inappropriatelyconsider number of trips as acontinuous variable at the household-level. Second, a variable thatcaptures the composite effect oftravel cost, in-vehicle time, and out-of-vehicle time by all availablemodes to all candidate traffic surveyzones was included as an accessibil-ity measure in trip productionmodeling.

Where are trips concentrated?The current research estimated

an individual-level destination choicemodel that appropriately combinesthe travel cost, in-vehicle time, andout-of vehicle time level of servicemeasures by all available modes todevelop a composite impedancemeasure for travel between each

zonal pair. The procedure alsorecognizes differential sensitivities tolevel of service measures acrossdifferent demographic groups in thepopulation.

How and when are the trips made?Many TCMs such as High-

Occupancy Vehicle (HOV) lanes,improved transit service during peakhours, peak period pricing, etc. arefocused on the peak periods. SuchTCMs have a non-uniform (acrosstime-of-day) effect on modal level ofservice. The primary effect of suchTCMs is to shift travel modes and/orshift time-of-day of travel. While theformer may be the predominantimpact for work trips, both mode andtime-of-day may be affected for non-work trips. This research developedmode and time-of-day models toanalyze TCM impacts on both thesedimensions.

Develop Models for SupplementaryTraffic Inputs Needed by MOBILE

As is the case of travel demandmodels, the development of modelsfor supplementary traffic inputs waspursued along several directions, asdiscussed below.

VMT Mix ModelsThe MOBILE model requires

several traffic related inputs, one ofwhich is the Vehicle Miles Traveled(VMT) mix ratio. Current proceduresdetermine the VMT mix ratio as afunction of two control variables:roadway type (freeways/arterials)and area type (rural/urban). It is quitelikely that there is substantialvariation in VMT mix even aftercontrolling for roadway functionalclassification and area type. Theresearch undertaken in this projectdeveloped a fractional split model

Project Summary Report 1838-S – 2 –

that determines the VMT mix ratio as afunction of several informative variables,including the physical attributes of links(such as number of lanes and whether thelink is a divided road or not), theoperating characteristics of links (such aslink speed), aggregate area-type character-izations of the traffic survey zone inwhich the link lies (such as urban,suburban, and rural), and the land-useattributes of the zone (such as retailacreage and manufacturing/warehouseacreage in the zone).

Travel Time Duration and Soak TimeModels

In addition to VMT, the MOBILEmodel takes several traffic-related data asinputs, one of which is the distribution ofthe duration of vehicle trips in the region.The vehicle trip duration distribution isimportant for several reasons. First, thetrip duration distribution providesinformation for developing trip durationactivity parameters used by the MOBILEemissions factor model to estimaterunning loss emissions. Running lossemissions are evaporative emissions thathave escaped from a vehicle while theengine is operating (from spots where thevehicle’s evaporative/purge system hasbecome inoperative). Second, operatingmode fractions, which are needed byMOBILE5 to estimate emissions rates,can be estimated from the trip durationdistribution. Third, the trip durationdistribution can be used to predict thevehicle miles of travel (VMT) accumu-lated on local roads in the region. Thecurrent research project formulated andimplemented a methodology for modelingtrip durations using vehicle trip data fromhousehold travel surveys and supplemen-tary zonal demographic/land-use data.

Another traffic input to the MOBILEmodel is the distribution of the soak-timeof vehicle trip starts. The U.S. Environ-mental Protection Agency (EPA) hasdefined soak-time as the duration of timein which the vehicle’s engine is notoperating and which precedes a successfulvehicle start. The soak-time for a trip islikely to depend on various factors suchas the activity purpose preceding the tripstart (i.e., the origin activity purpose), thetime-of-day of the trip start, and otherland-use and socio-demographic charac-teristics of the zone of trip start. Thisresearch formulated and implemented amethodology for modeling soak-timedurations. The methodology involvedestimation of models using vehicle tripdata from household travel surveys and

supplementary zonal demographic/land-use data. The effectiveness of themethodology lies in its easy application atthe traffic zonal level within a metropoli-tan region to obtain zone-specific soak-time distributions by time-of-day andorigin activity purpose.

Integrate Relevant Models within aGeographic Information System (GIS)Architecture

In addition to refining the currentregional travel models and developingmodels for improved supplementarytraffic inputs to emissions models, thisresearch focused on their integrationwithin TransCAD, a premier GIS fortransportation planning. TransCAD isideally suited for transportation air qualityanalysis because such an analysis isintrinsically spatial and requires thestorage and manipulation of vast amountsof spatial data. The current researchdeveloped graphical user interfaces in theTransCAD environment to implement thetravel demand and traffic input modelsdescribed earlier. These user interfaces areeasy to use and guide the user through themodeling process using dialog boxes andprompt windows. The research alsointegrated the outputs from various traveldemand models with a comprehensivemap display of the DFW region. This willenable the user to visualize the modelresults.

What We Found...There are five broad findings from

this research.First, sociodemographic and

employment related attributes have asignificant impact on travel-relateddecisions of households and individuals.While this result is not surprising, thestate of the practice in travel demandmodeling continues to use very limitedspecifications of sociodemographicvariables in the modeling process. Suchlimited demographic variable specifica-tions can, and in general will, lead tomisinformed transportation planning andair quality decisions because of projectedchanges in demographic and employment-related trends over the next few decades(including aging of the population, adecrease in the number of householdswith children, and more employedindividuals). In addition, our resultsindicate differential sensitivities ofsociodemographic groups to transporta-tion system performance. Accommodatingsuch differential travel sensitivities ofdemographic groups is important foraccurate evaluation of transportationcontrol measure as well as for environ-mental justice considerations in transportpolicy analysis.

Second, it is practical and veryfeasible to apply models in forecasting

Figure One : Display of Trip Interchanges by Mode and DepartureTime for the DFW Metropolitan Area

Project Summary Report 1838-S – 3 –

.

mode even when they involve severalexplanatory demographic variables. Oneof the reasons often provided for the useof a limited specification of demographicvariables in estimation is that it rendersthe forecasting process manageable.However, this research project hasdeveloped and applied a forecastingmethodology that can be integrated withina GIS-based platform, even if severalexplanatory variables are used duringestimation. The methodology, which israther simple, entails the generation of asynthetic population of households andindividuals based on current or projectedaggregate population demographicdistributions.

Third, localized transportationcontrol measures (TCMs) such aswidening existing lanes, adding newlanes, and traffic signal improvementsmay not substantially improve mobilityand emissions levels. This finding hassignificant implications for the choice ofTCMs for air quality improvements, sinceit suggests that corridor or area wideTCMs such as HOV lanes, car poolingincentives, and alternative work arrange-ments may be more effective thanlocalized TCMs in improving mobilityand air quality. However, because of thelimited nature of the simulation exercisesin the current project, we recommendcaution in generalizing this finding.Specifically, the results may be areflection of the relatively low to mediumlevels of congestion (as opposed to veryhigh levels of congestion) on the arterialsused in the simulation exercises. A morecomprehensive examination of localTCMs in the context of different roadwaytypes, different vehicle mixes, anddifferent volume levels would bedesirable in making any strong andgeneral conclusions. Notwithstanding thecaveat above, the results do question thecurrent practice of assuming blanketemissions benefits of localized TCMs.

Fourth, current approaches to VMTmix, local road VMT, soak time distribu-tion, and VMT by trip time duration canbe substantially improved by developingmodels based on local vehicle classifica-tion counts and survey data. Since theemissions computations in the MOBILEmodel are very sensitive to these inputs, itis important that metropolitan planningorganizations consider pursuing suchefforts.

Fifth, visualization of the intermedi-ate and final traffic output resultsgraphically on the Texas network aids inunderstanding traffic patterns and

provides an effective intuitive means tochecking model functionality.

The ResearchersRecommend...

Our recommendations are providedunder two categories: Implementation andFurther Research.

Implementation RecommendationsThe models developed as part of the

integrated transportation air qualityprocedure can be implemented for theNorth Central Texas Council of Govern-ments (NCTCOG) area. The models fortravel demand will need additional effortto ensure compatibility with the currentframework adopted by NCTCOG;however, the models for supplementarytraffic inputs can be used as they are byNCTCOG for planning purposes.Implementation of the travel demand andsupplementary traffic model formulationsfor other metropolitan areas will requiremodel estimations based on data collectedlocally. It is recommended that TxDOTpursue such implementation-related workfor other Texas metropolitan areas.

Research RecommendationsThe U.S. Environmental Protection

Agency (EPA) is currently developing theMOBILE6 emissions factor model, whichshould become the required standard for

air quality conformity and transportationcontrol measure (TCM) effectivenessanalysis by Fall 2001. The structure aswell as the input needs for MOBILE6 arequite different from those of its predeces-sor MOBILE5. The changes in MOBILE6are valuable and should result in moreaccurate mobile source emissionsestimates. At the same time, the MO-BILE6 model offers substantial opportu-nities and poses important challenges toimprove traffic inputs. Among theseinputs are (a) fleet characterization data(projections of future vehicle fleet size;and fraction of travel by a multidimen-sional breakdown based on vehicle age,mileage accumulation rate, and thirtyvehicle types), (b) separate traffic-relatedvariables for weekdays and weekends inemissions modeling, and (c) a very hightemporal resolution during the day for alltraffic indicators. It is recommended thatTxDOT place a high priority on thedevelopment of models that are capable ofaccurately providing such traffic inputs tothe MOBILE6 model.

Figure Two: Display of the VMT Mix on an IH-45 SB ramp inDallas, Texas.

Project Summary Report 1838-S – 4 –

The University of Texas at AustinCenter for Transportion Research3208 Red River, Suite #200Austin, TX 78705-2650

Disclaimer

Research Supervisor: Chandra R. Bhat, Ph.D., (512) 471-4535email: bhat@mail.utexas.edu

TxDOT Project Director: William Knowles, (512) 465-7648email: wknowle@dot.state.tx.us

The research is documented in the following reports:1838-1 Description of Data Acquisition Efforts, August 19991838-2 Review of 1990 Mobile Source Emissions Modeling Procedure for the Dallas-Ft. Worth Non-Attainment Area, August 19991838-3 Review of the Current Dallas-Ft. Worth Regional Travel Demand Model, August 19991838-4 Presentation Slides for Review Meeting on August 19,1999, September 19991838-5 VMT Mix Modeling for Mobile Source Emissions Forecasting: Formulation and Empirical Application, May 20001838-6 Modeling Soak-Time Distribution of Trips for Mobile Source Emissions Forecasting: Techniques and Applications, August 20001838-7 Modeling Trip Duration for Mobile Source Emissions Forecasting, August 20001838-8 Transportation Control Effectiveness in Ozone Non-Attainment Areas, August 2001

To obtain copies of a report: CTR Library, Center for Transportation Research,(512) 232-3138, email: ctrlib@uts.cc.utexas.edu

The research resulted in the development of a GIS-based travel demand model that is environmentally policysensitive and provides improved assessments of Transportation Control Measure (TCM) effectiveness in ozonenonattainment areas. The model is imbedded in TRANSCAD software and is used to estimate mobility andemission changes due to TCM Implementation. The model will be used by Metropolitan Planning Organizationsand TxDOT to streamline the transportation conformity process.

For more information please contact Bill Knowles, P.E., RTI Research Engineer at (512) 465-7648 orwknowle@dot.state.tx.us.

This research was performed in cooperation with the Texas Department of Transportation and the U. S.Department of Transportation, Federal Highway Administration. The content of this report reflects the views of theauthors, who are responsible for the facts and accuracy of the data presented herein. The contents do not necessarilyreflect the official view or policies of the FHWA or TxDOT. This report does not constitute a standard, specification,or regulation, nor is it intended for construction, bidding, or permit purposes. Trade names were used solely forinformation and not for product endorsement. The engineer in charge was Chandra R. Bhat (Texas No. 88971).

Your Involvement Is Welcome!

For More Details...

TxDOT Implementation StatusApril 2002

PRSRT STDUS Postage PaidPermit No. 391Austin, TexasThe University of Texas at Austin

C e n t e r f o rTransportation Research