TMIPConnection
Issue 8, July 1998

The Travel Model Improvement Program Newsletter

The Travel Model Improvement Program is sponsored by:
U.S. Department of Transportation
Federal Highway Administration
Federal Transit Administration

Table of Contents

New TMIP Review Panel Members

John Simons and Joseph Schofer have agreed to join the Program Review Panel. Mr. Simons is with the Taubman Company, a major national developer. In a past life he was a traffic engineer with Barton-Aschmann Associates.

Dr. Schofer is a professor of Civil and Transportation Engineering at Northwestern University where he is chairman of the Department of Civil Engineering. He has also done research covering a broad range of transportation topics.

Table of Contents

TEA-21 Authorizes TRANSIMS Deployment

The Transportation Equity Act for the Twenty First Century (TEA21), passed in May of 1998, authorizes the deployment of TRANSIMS. Section 1210, Advanced Travel Forecasting Procedures, provides funds for the completion of the TRANSIMS core development, packaging in a user friendly format, training and technical assistance for users. Beginning in the year 2000, the bill also provides financial support on a cost sharing basis for a limited number of urban areas to convert from existing forecasting procedures to TRANSIMS. TEA21 allocates $25 million to this effort over six years, from 1998 to 2003. Currently the TMIP staff is developing plans to implement this section of the Bill and will provide further information as these plans are finalized.

Also of interest to planners is the Transportation and Community and System Preservation Pilot Program. This program supports local areas in preserving communities and developing and implementing sustainable transportation plans and policies. The program provides for planning grants, implementation grants and a research effort to address related issues. A more detailed description of the program will appear in the Federal register.

Table of Contents

Getting Ready for TRANSIMS

By Ken Cervenka and Mahmoud Ahmadi, North Central Texas Council of Governments

When you hear the word TRANSIMS, what thoughts come to YOUR mind? Over the past few years, we have heard opinions about the TRansportation ANalysis SIMulation System ranging from "It is impossible to implement" to "It is not detailed/complex enough." Many MPO staff are adopting a "let's wait and see" attitude, in which they want solid proof of value before taking TRANSIMS seriously. This conservative view is understandable, for an individual-level simulation of regional travel falls outside the "comfort zone" of today's travel model practitioners.

Back in 1994, Michael Morris (the Dallas-Fort Worth MPO director at NCTCOG) summarized TRANSIMS as having "the highest possible gain, as well as the highest possible risk." This statement was made during the first visit to NCTCOG by Los Alamos National Laboratory (LANL) staff, in which we learned about the proposed TRANSIMS framework and LANL learned about the kinds of work an MPO does--as well as our interest in "next generation" travel forecasting/analysis tools. It is now four years later, with the Dallas Case Study completed in 1997 and the Portland Case Study under development, and Michael's statement still rings true. A full TRANSIMS implementation will represent a remarkable step forward in best-practice procedures, but there are no guarantees as to when this will take place or the level of success that will be achieved.

NCTCOG's role in the TRANSIMS project has been to provide LANL with network and travel data for TRANSIMS testing, as well as to review output and offer suggestions for improvements. TRANSIMS Release 1.0 was installed earlier this year on NCTCOG's Sun UNIX computers, and we now have an opportunity to see for ourselves what can be done with this interim package. [Note: the TRANSIMS User Notebook, prepared March 1998, contains over 350 pages!]. Considerable sensitivity testing and evaluation of the new tools remains to be done--much more than we can possibly do ourselves, which is why we want universities and consultants to actively participate in TRANSIMS testing and evaluation.

If you work for an MPO, is there anything you should be doing, right now, to get ready for TRANSIMS? Here are a few suggestions:

  • Buy the Computing Power You Need, But Only For Your Current/Near-Term Activities. The computational speed for a full-scale regional microsimulation will always be a subject for discussion. However, the price/performance ratio for computers continues to drop so dramatically that it would be unwise to purchase massive computing power until you have a real near-term need for it. While the current TRANSIMS software runs only in the UNIX environment, you shouldn't let this influence your agency's computer operating system decisions until you are much closer to an actual TRANSIMS implementation schedule.
  • Continue Your Ongoing Model Improvement Activities. One of the more unusual "criticisms" of TRANSIMS is that, if this is to eventually replace current travel forecasting/analysis activities, why bother with continued improvement of these other procedures? NCTCOG's position is that MPOs should always strive to implement "best practice" models--since TRANSIMS is not yet ready for wide-spread implementation, current model improvement activities should be centered around more traditional procedures. Any future transition from a good travel model system to TRANSIMS will be easier (and probably more successful) than a transition from a poor or non-existent travel model system.
  • Improve Your Current Information Systems. A term coined in the 1960s, with the advent of increased computer usage, is GIGO: Garbage In, Garbage Out. Regardless of whether TRANSIMS is ever implemented in your region, your agency could probably benefit from an improved data management system. Investment in "clean up" of demographic (land use), travel model network, travel survey, and observed travel data should be of value to any travel forecasting or information dissemination process. As one technique to control the GIGO problem, we are implementing, here at NCTCOG, a stronger integration of our databases and travel modeling process with GIS-based procedures.
  • Get Some Traffic Engineering Expertise On Board. One of the interesting side effects of TRANSIMS is that it will force travel modelers to consider traffic operations impacts at a greater level of detail than simple link-based volume/capacity ratios. Since MPO activities often already include evaluations of Congestion Management System, Intelligent Transportation System, and Transportation Control Measure strategies, this need for a knowledge of traffic engineering principles should already be evident.

We still remain optimistic about TRANSIMS' promise to improve travel forecasting/analysis procedures and, ultimately, the entire transportation decision-making process. LANL has been charged with a very difficult task, which is to find a suitable balance of their resources among the requirements for basic research, software development, and actual implementation/application.

Thank you for your efforts, so far!

Table of Contents

Availability of Synthetic Population Module

By Dennis Perkinson, Texas Transportation Institute

The role of household demo- graphics is central to trip generation in traditional travel demand forecasting. The traditional travel demand forecasting process typically involves preparation of base year estimates of population, number of households, average house-hold size, median household income, and employment at the travel analysis zone (TAZ) level. Trip generation rates are then estimated for each TAZ as a function of these variables (e.g., trip rates stratified by household size and income levels), using some statistical method (e.g., simple cross-classification, regression, etc.). Thus, in the traditional travel demand forecasting process, trips move between zones as a function of the characteristics of those zones, which are in turn generated as a function of the aggregate house-hold characteristics within the zones. (Trips are attracted to zones as a function of other characteristics, such as employment.)

The traditional four step travel forecasting process does not incorporate time of day. Models are typically run for an aggregate 24 hour day or the peak period. Consequently, shifts in time of travel as encouraged by Transportation Control Measures (TCMs) are not captured by the traditional process. The traditional process cannot readily analyze changes in peak period behavior, such as peak spreading and particularly congestion pricing.

The Transportation Analysis Simulation System (TRANSIMS) is a disaggregated simulation-based model of urban transportation demand. TRANSIMS requires individual travelers with individual itineraries defined by origin, destination, time of day, and trip purpose. Individual travelers must come from individual households rather than aggregated travel analysis zones. The elimination of zonal aggregation in trip generation begins with individual household demographic data.

An exhaustive survey (i.e., all households) of an area is infeasible economically and logistically. The challenge is to create a database of individual household characteristics for all households (i.e., not a sample or summary statistic) from available sources. As part of the development of TRANSIMS, Los Alamos National Laboratory has developed procedures and software designed to produce populations (e.g., family households, non-family households, and groups quarters) that are statistically equivalent to actual populations (at the Census block group level).

Very briefly, the synthetic population generating procedure uses summary tables based on a representative sample and data on representative individuals to produce the individual level of detail required by TRANSIMS. The Census provides two data files which contain the desired information. These are the Standard Tape File 3A (STF-3A) and the Public Use Microdata Sample (PUMS). The technical details and logic of this procedure are described in detail in Beckman et al. 1996.

The procedure described above generates baseline (current) household demographics. A procedure for forecasting future disaggregated household demographic data from baseline disaggregated household data is also being developed. Ultimately household demographic characteristics will be linked to activities, which in turn will be linked to trip origin, destination, time-of-day, and purpose, as well as being the basis for forecasting future household characteristics and activities.

The current (Unix based) version of the Synthetic Population System is in the process of being converted to operate in a Windows environment. A detailed user's guide is in preparation. Both the Windows conversion and the enhanced user's guide are expected to be available for field testing later this year.

Table of Contents

TRANSIMS Travelogue – July 1998

TRANSIMS TRAVELOGUE describes current activities within the TRANSIMS project.

What is TRANSIMS?

The TRansportation ANalysis and SIMulation System (TRANSIMS) is one part of the multi-track Travel Model Improvement Program sponsored by the U.S. Department of Transportation, the Environmental Protection Agency, and the Department of Energy. Los Alamos National Laboratory is leading this major effort to develop new, integrated transportation and air quality forecasting procedures necessary to satisfy the Intermodal Surface Transportation Efficiency Act and the Clean Air Act and its amendments.

TRANSIMS is a set of integrated analytical and simulation models and supporting data bases. The TRANSIMS methods deal with individual behavioral units and proceed through several steps to estimate travel. TRANSIMS predicts trips for individual households, residents and vehicles rather than for zonal aggregations of households. TRANSIMS also predicts the movement of individual freight loads. A regional microsimulation executes the generated trips on the transportation network, modeling the individual vehicle interactions and predicting the transportation system performance. Motor vehicle emissions are estimated using traffic information produced by TRANSIMS.

Project Approach

We are developing two interim operational capabilities (IOC) to cover the major TRANSIMS components: Household and Commercial Activity Disaggregation, Intermodal Route Planner, Transportation Microsimulation, and Environment (primarily vehicle emissions). As each IOC is ready and with the collaboration of a selected metropolitan planning organization (MPO), we will complete a specific case study to confirm the IOC features, applicability, and readiness. This approach should provide timely interaction and feedback from the TRANSIMS user community and interim products, capabilities, and applications.

Traffic microsimulation was emphasized in the first IOC, which we tested in a case study in the Dallas-Fort Worth region with the support of the selected MPO, the North Central Texas Council of Governments. The next IOC focuses primarily on the iteration among the Activity Demand, Intermodal Route Planner and the Transportation Microsimulation, but also will incorporate the capability for emissions predictions. The supporting case study will be situated in Portland, OR, where we are working with Portland METRO.

President Clinton Sees TRANSIMS

On February 3, 1998, William J. Clinton, President of the United States, was shown a computer-generated animation of the results of the TRANSIMS Dallas-Fort Worth case study by Dr. John Browne, Los Alamos National Laboratory Director. The President visited Los Alamos to give a speech supporting the scientific research and technology necessary to ensure the safety and reliability of the nation's nuclear stockpile. He also acknowledged the Laboratory's contributions to meaningful advances to other pressing national security issues, such as environmental clean up and reducing greenhouse gas emissions. The TRANSIMS demonstration was one of only three technologies presented by the Director to the President and Federico Pena, Secretary of Energy and former Secretary of Transportation.

IOC-2 Planning

We have begun initial planning for the case study that will demonstrate the capabilities of the second IOC. Although our plans are not yet firm, we are considering a study that involves activity demand forecast, intermodal trip plans, and microsimulation for all of the Portland METRO planning region. We also are considering a forecast scenario for a future year. We expect to include the most travel modes (auto, bus, light rail, walk, and bicycle) in the study, and a high occupancy vehicle analysis is receiving strong consideration. Trucks and freight also will be represented. The environmental analysis will involve emissions only, not air quality.

Portland METRO has provided us with an EMME/2 representation of the Portland transportation network as a starting point for our research and development of the methods for IOC-2. This EMME/2 representation has been augmented with additional features required by TRANSIMS such as pocket lanes, signals, and lane connectivities at intersections. Portland METRO also has begun generation of a detailed network for the base year transportation system. The detailed network starts with Tiger files and will be transformed into the TRANSIMS format. We have named this the "all streets" network. In addition, we have revised the TRANSIMS network representation to accommodate multi-mode transportation.

In the following sections, we report our ongoing efforts for extending our microsimulation and planner methods to multiple modes. We contracted with the National Institute of Statistical Sciences to adapt their activity forecasting methods for implementation within the TRANSIMS framework, and we are investigating an alternative activity forecasting method with Portland METRO.

Microsimulation Enhancements

The first TRANSIMS IOC was a vehicle-based simulation. For the next IOC, which considers multiple modes, we have begun to implement a traveler-based simulation. Travelers are persistent objects that must be tracked through the simulation. Car drivers and passengers must be matched with specific vehicles; transit passengers with specific routes. Drivers can no longer create vehicles, but instead must have a vehicle at a parking accessory before they can begin a trip. The vehicles they use must obey constraints such as remaining where they are parked and not become available elsewhere in the simulation. Travelers cannot appear in more than one place at a time in the simulation.

We added support to the TRANSIMS cellular automata microsimulation for transit vehicles. They follow assigned routes and stop at specific locations to pick up and discharge passengers. We added the ability to simulate multicell vehicles (i.e., longer than 7.5 meters) with vehicle-specific accelerations, maximum velocities, and capacities. We added car-pooling and for non-microsimulated transportation modes, such as walking. We track every individual second by second in the simulation.

Planner Methods

We are developing methods for formal language constrained shortest paths on a hierarchical network for use in the TRANSIMS Intermodal Route Planner. In these methods we create separate networks for each transportation mode. For instance, we may have separate networks for automobiles, buses, walking, bicycles, light rail, etc. Furthermore, within the transit networks we represent each route as a subnetwork.

The perceived or real costs of a link's traversal are assigned to each link in these networks. Where a mode change can occur, we have a transfer link between the two different mode networks. This transfer link also has the associated intermodal transfer cost whether it be time, money, and other attributes or qualities that may influence a travelers mode decision.

Each link on a mode network also is assigned a label unique to the mode. For instance, the automobile network may be labeled 'c'; the bus network,'b'; the walk network, 'w'; the rail network, 't'; etc. Then we establish a formal language that describes the sequence of labeled links, identifying mode choices, for paths through the network. For example, an all-walk path is described by w+, where w+ denotes a sequence of walk links. A walk to the bus stop, followed by riding the bus to a downtown stop, and then completing the trip with a walk to the work place is denoted by w+b+w+. Some paths are not allowed--for example, b+c+b+c+ would be an unusual path for a commuter, requiring having an automobile stashed, and requiring jumping from bus to car and car to bus.

Thus we establish a set of allowed sequences within the formal language and assign probabilities for using each sequence. Initially the probabilities may be uniform. They may be conditional on the household demographics; for example, if a car is unavailable, the chances of c+ being in the sequence are small. In our research, we have developed fast-running algorithms that find minimum cost paths having these labeled sequences through the multi-modal network and thus generate trip plans for travelers. After the microsimulation has executed the trips, in addition to updating the link costs, the information will be fed back to the planner to adjust the probabilities for picking the mode sequences. Mode splits thus can emerge from the iteration between the planner and the microsimulation. The technique has the advantage that when a forecasted transportation system or policy is simulated, the mode split depends on the system and policies.

We also are investigating path finding algorithms that find the closest destination from among several destinations when a traveler has several alternative locations for accomplishing his activities. Thus we are researching efficient interfaces between the route planner and the activity generation module, focusing on efficient ways to encode mode preferences and choices that depend on the order of performed activities. We are examining efficient algorithms for time-dependent, mode choice constraint shortest paths and routes, route problems with time window constraints and related scheduling and routing problems.

Activity Demand Methods

We are pursuing three methodologies for generating activities for the population. The first is an interim methodology that only assigns work trips to the workers in the population. The trips are assigned by using both census and land use data. We generated a set of such activities using the population to identify the workers and the home location, and the land use to identify the work location. These activities are being refined as morning peak traffic conditions are studied and its effect on the morning work activity location is understood.

These activities are input to the router/planner to facilitate router/planner research. The activities generated at Los Alamos represent a base set that will be modified using feedback from the router/planner and the microsimulation. These activities are assigned to the synthetic households that were generated for Portland and placed on the EMME/2 network. This population will be placed on the "all streets" network, and activities likewise will be assigned to the links on the "all streets" network.

The National Institute of Statistical Sciences (NISS) is developing a second methodology. In this methodology they develop activities by resampling the Portland activity survey. Classification and regression trees are used to assign activity sets to complete households. The activities in the activity set then are assigned to individuals in the household. The activities are located by fitting a density function to the activity locations in the Portland survey. Land use surrounding the network links is used to refine the activity location. Mode choice is not available at this time in this methodology, but the methodology used by Mark Bradley could be employed.

Under contract to Portland METRO, Mark Bradley has generated a third set of activities. Using a nested logit approach, each person in the traveling population is given activities. The activity locations are based on characteristics of Traffic Analysis Zones (TAZ). The activity locations are assigned at random within the TAZ. This will change as land use in the vicinity of the links will be used to assign activity locations. Mode choice also is given in the activity set. It was determined by a logit fit to the Portland activity survey data.

As development of the latter two methods progresses, they will be continually evaluated for incorporation into the TRANSIMS methodology. The activity demand methods will have to be compatible with the feedback methods that use information from the router/planner and the microsimulation to adjust activity sequences, times and locations. Another important consideration will be how well the activity demand methods represent the sensitivity of individual and household activities to changes in the transportation infrastructure and policies. If both methods fit well within the TRANSIMS methodology and time and resources permit, we may compare the results from the two methods as part of the Portland case study.

Further Information

For further information about the TRANSIMS program, please contact:

Dr. LaRon Smith
Los Alamos National Laboratory
P.O. Box 1663
Mail Stop F606
Los Alamos, New Mexico 87545
Phone: (505) 665-1286
Fax: (505) 665-5249
E-mail: llsmith@lanl.gov

Table of Contents

Portland Case Study

By Keith Lawton, Portland METRO

The activity model originally developed for METRO is being modified to make it suitable for TRANSIMS.

The daily activity model developed for METRO was limited to five rather broad time periods. While it was a microsimulation (sample enumeration) on synthetic households, it was not stochastic in the sense of determining the specific (unique) pattern chosen. Instead it assigned probabilities to pattern, time of day, mode and destination choice. The patterns were limited tours which had no more than 5 activity stops, including a work-based sub tour. The tours were also limited to three primary purposes (work/school, household maintenance and discretionary).

The project consultant on the METRO model conversion has made considerable progress in enriching and greatly amending the activity scheduling model for TRANSIMS application.

The basic METRO style model has been changed from a probability model to stochastic using Monte Carlo simulation. When applied to a 100% count of synthetic households, this results in a full set of activity patterns and tours limited to the METRO classification by primary purpose, complexity and time period designation.

The activity patterns/tours available in the 1994/1995 METRO household survey (which have discrete times within the time periods used and are specific as to individual purposes and also have all stops on each recorded tour) are then categorized by the METRO model classification, and sampled randomly within classification to provide discrete tours which have full information. In short, the result is a synthetic household survey for all households in the METRO region!

A prototype model (in Pascal) has been written which accomplishes steps 1 and 2.

Conversion of this prototype to C is underway.

Still to come are the addition of the under 16 age group travelers to the Model and the separation of work and school "purposes". Also they are moving away from traffic analysis zones to discrete lots, or street arcs.

METRO will use this advance to stochastic determination and to a more discrete time of day grouping for application at METRO prior to the full availability of the TRANSIMS model suite. This will yield a very rich set of output data for policy analysis.

A major near term effort will be to calibrate the new model to known secondary data (traffic counts, transit counts etc.). The survey (as is typical in a home interview survey) is biased towards less mobile households. Calibration of a model yielding activity patterns arranged in complex tours will be quite challenging. Model re-estimation will likely be needed if mis-specification becomes evident.

Table of Contents

Portland TRANSIMS Network Update

METRO's Travel Forecasting Section is assembling information about the Portland-Vancouver region's current transportation infrastructure for the next TRANSIMS case study.

METRO's enhanced TIGER file for the four-county region, including local streets and edited to facilitate routing, forms the basis of the TRANSIMS network. Custom GIS interface tools have been built to assist data processing and attach attributes to network nodes or to the appropriate side of network links.

Several sets of network data have already been prepared for Los Alamos National Laboratory. Speed limit, functional class, capacity, allowed modes, and prohibited turns were transferred from METRO's EMME/2 modeling network. Four-foot aerial photographs were inspected for allowed on-street parking, the number of thru lanes, and the length of turn pockets and two-way left turn lanes. USGS soil surface elevation figures at network nodes were edited to account for street alignments above or below the surface. Traffic signal locations were collected from cities, counties, and state DOTs. Transit stops and a typical run for each transit route were attached to the network based on data provided by the region's transit agencies.

Efforts are currently underway to enhance the existing network data. Off-street and on-street parking capacities are being estimated from a combination of direct counts and factors developed by an earlier METRO parking study. Scheduled daily itineraries for every transit vehicle are being attached to the network. A regional sign inventory is being tapped to locate stop, yield, and turn signs. Electronic timing and phasing plans for signalized intersections are being collected from the signal-operating jurisdictions. Speed limits and functional class are being updated for all links based on data collected from jurisdictions.

It is anticipated that the first round of data transmissions will be complete in fall 1998. METRO staff will then coordinate with LANL in producing a data preparation guide for MPO implementation of TRANSIMS. In the interim, any specific questions about METRO's TRANSIMS network data may be referred to:

Bill Stein, Associate Transportation Planner
METRO Travel Forecasting Section
600 NE Grand Avenue
Portland OR 97232-2736
(503) 797-1855

Table of Contents

New TMIP Publications

Nonresponse in Household Travel Surveys, DOT-T-98-4

This report is written for designers, analysts, sponsors of household travel surveys, and all other persons who find themselves involved, in one way or another, with the collection, reporting, or interpretation of travel survey data. Its objective is to provide a set of guidelines for measuring and reporting nonresponse in household travel surveys and for reducing the level and impact of non-response. To accomplish these goals, this report used a three-pronged approach as described in the chapter summaries below.

Chapter 1. Measuring and reporting nonresponse. A standard approach to reporting response rates is recommended. A standard approach, used consistently, can help assess the quality of survey data. Standard reporting also allows users to evaluate different techniques for implementing surveys, thus building a coherent body of knowledge on methods for household travel surveys.

Chapter 2. Reducing nonresponse: To reduce nonresponse, characteristics of respondents and interviewers must be understood. Characteristics of typical nonrespondents to travel surveys are discussed. Procedures to improve response rates are recommended.

Chapter 3. Statistical methods for reducing the impact of nonresponse: Despite our best efforts, all travel surveys are likely to have nonrespondents. Specific methods to adjust survey results to better represent the population are recommended.

Introduction to Panel Surveys in Transportation Studies, DOT-T-98-3

This report provides a general introduction to the use of panel designs in surveys of travel behavior. It has four main objectives:

  1. to highlight the differences between cross-sectional and panel approaches to the study of travel behavior,
  2. to discuss the limitations of cross-sectional and panel data,
  3. to identify situations where panel data are preferable, and
  4. to provide guidelines for designing and maintaining a panel survey.

Through examples drawn from the transportation literature, this report illustrates how panel designs can be used to address a variety of transportation issues. The report identifies several situations where panel designs are preferable, either because they provide information that cannot be obtained on cross-sectional designs or because they are more efficient than cross-sectional designs. It then discusses the special issues and problems that arise when the same group of individuals is followed over time. The final sections of the report provide guidelines for designing and maintaining a panel survey, and for preparing panel weights for analysis of the data.

Travel Model Speed Estimation and Post Processing Methods for Air Quality Analysis, DOT-T-98-5

Transportation planners have relatively sophisticated and complex computer models available to them for forecasting travel demand and air quality. The weak point in the process however is the interface between the demand and pollutant emission models. Travel demand models are designed to forecast travel demand but have not traditionally been as reliable for forecasting vehicular speeds. Air pollutant models however require as input reliable estimates for vehicle demand, vehicle speeds, and vehicle operating mode (e.g. cold start, hot start, etc.). This gap between the traditional outputs of travel demand models and the required inputs of air quality models is the subject of this report.

This report suggests various short term improvements that might be made to the speed estimation routines contained in travel demand models, and suggests various post-processor routines that can be used to further improve model speed estimates. These post-processor routines generally use data and proced- ures not typically available in travel demand models. Finally this report suggests improve-ments that can be made in current techniques for estimating vehicle operating modes.

Activity-Based Travel Forecasting Conference, DOT-T-97-17

A report on a conference to promote use of activity-based approaches for travel forecasting. Corollary purposes were to identify activity-based forecasting techniques that can be used now and to recommend actions to advance the state-of-the-art. The conference was organized as one plenary session and three workshops. This report includes papers that document the introductory seminars, the keynote address, and five other presentations in the plenary session. Also included are summaries of the discussions and recommendations in the three workshops:

  • Data Resources and Survey Methods for Activity Analysis
  • Models of Activity and Travel Behavior
  • Microsimulation in Activity Analysis

Quick Response Freight Manual, DOT-T-97-10

The manual addresses freight issues at different levels of analysis. On the more detailed site planning level, the methods include predicting the number and temporal distribution of truck trips to and from specific locations and identifying the routes used. On a more aggregate level such as corridor, metropolitan area, or regional level, the manual helps develop forecasts of trips generated by various traffic analysis zones and distribute these trips to the transportation network. The analytical methods contained in the manual place special emphasis on inclusion of transfer-able parameters that can be used as default values for model inputs when data specific to the State or metropolitan area are not available. This manual also identifies alternative analytical methodologies and data collection techniques in order to improve the accuracy of the freight analysis and planning processes.

Transfer Penalties in Urban Mode Choice Modeling, DOT-T-97-18

This is a report on recent research done on the subject of transfer penalties in urban mode choice decision-making. The research was undertaken to determine whether such penalties, thought by many observers to exist, are quantifiable. If so, the results provide a firmer basis on which to gauge the merits of travel model sets and the forecasts of transit ridership that they produce. Using Boston area household travel survey data and hand-coded transit impedance data, mode choice models containing transfer-related variables were estimated by the Central Transportation Planning Staff. Transfer penalties for work trips were, in fact, found and quantified. In models based on an extremely carefully construct- ed data base, a transfer dummy variable representing whether a transfer is required to complete a trip, emerged as a modestly significant variable. It was found to be equivalent to 12 to 15 minutes of transit in-vehicle time, depending on the particular model specification. Other interesting and useful information emerged from this research as well. It was found that hand-coding transit impedances and establishing fairly liberal definitions of whether transit is an available mode had a significant impact on model estimation and resulted in more accurate parameter coefficients.

One free copy of these reports can be obtained by writing US DOT, Subsequent Distribution Center, Ardmore East Business Center, 3341 Q 75th Ave., Landover, MD 20785. Please include the complete report title, the report number, and your complete mailing address. Information on additional TMIP reports or the TMIP Program is available from Joe Kammerman at (202) 366-4054 or joseph.kammerman@fhwa.dot.gov.

Table of Contents

TMIP reports are now available on the net!

Beginning this month, an order form will be available on the net. Having order forms on the world wide web will enable you to get fascinating facts and engaging publications even faster than the old way (writing a request for an order form, waiting to receive the order form, sending the order form in, and then waiting to receive your publications). The order form can be found on the TMIP web site. Please utilize this new service.

Table of Contents

TMIP E-Mail List

TMIP now has its own e-mail list (sometimes referred to as a listserv or mailing list). This list has been set up as an open forum for transportation professionals to discuss relevant issues. Recent topics have included:

  1. Well, what *is* a place-based survey?
  2. TEA-21 and Travel Forecasting Programs
  3. TRANSIMS in Europe
  4. Nonrecurring Congestion Model/Magnitude
  5. Incident Delays
  6. Observed Elasticities

It is free and open to anyone who is interested in the transportation profession. To subscribe to the TMIP E-Mail List, send a message to subscribe@list.bts.gov with the following on the first line in the body of the message:

subscribe tmip <your name>

For example, you would type: subscribe tmip John Doe. You will be sent a welcome message and information about sending messages to the list, unsubscribing and how to change your list settings. You can also find subscription information on the TMIP Website. If you have questions about the TMIP E-Mail list, please send a message to Lisa Day (lday@tamu.edu).

Table of Contents

What's New on the TMIP Website?

The following are some recent additions to the TMIP Website. Be sure to check this location often for updates!

  • Review Panel Functions and Organization, SG Associates, Inc., 1994 [added 6/4/98]
  • Workshop on Transportation Air Quality Analysis, U.S. Department of Transportation, Federal Highway Administration, NHI Course No. 15265, September 1994 [added 6/3/98]
  • Transportation Management for Clean Air & Efficient Growth: On the Road to Progress?, Michael A. Replogle, Environmental Defense Fund [added 5/21/98]
  • Urban Transportation Planning in the United States - An Historical Overview, Fifth Edition, Ed Weiner, September 1997 [added 5/19/98]
  • Model Validation and Reasonableness Checking Manual, Barton-Aschman Associates, Inc. and Cambridge Systematics, Inc., February 1997 [added 5/14/98]
  • A Simulation Laboratory for Evaluation of Dynamic Traffic Management Systems, Qi Yang, Center for Transportation Studies, Massachusetts Institute of Technology, February, 1997 (revised version of Qi Yang's Ph.D. dissertation) [link added 4/22/98]
  • Transportation Conformity and Demand Management: Vital Strategies for Clean Air Attainment, by Michael Replogle, Environmental Defense Fund, April 30, 1993 Edition (with October 1, 1993 Supplementary Appendices) [added 4/8/98]

Table of Contents

Upcoming Conferences

  • APTA Annual Meeting, October 4-8, 1998, New York Hilton, New York, NY-- Sponsored by the American Public Transit Association. Contact Ingrid Tomasek (itomasek@apta.com) for further information.
  • Statewide Travel Demand Forecasting Conference, December 6-8, 1998, Beckman Center, Irvine, CA Sponsored by the Transportation Research Board and the Travel Model Improvement Program. Contact Jim Scott (jscott@nas.edu), 202-334-2965 or visit the TMIP Website for further information.
  • International Conference on Computational Intelligence for Modeling, Control and Automation - CIMCA '99, February 17-19, 1999, Vienna, Austria. Contact Masoud Mohammadian (cimca99@fcit.monash.edu.au), 61 351226136, Fax: 61 351226842 for further information.
  • Seventh Conference on the Application of Transportation Planning Methods, March 7-11, 1999, Boston, Massachusetts -- Sponsored by the Transportation Research Board. Contact Richard S. Marshment (rmarshment@ou.edu), 405-325-2399, Fax: 405-325-7558, Rick Donnelly 505-883-0029 or the TMIP Website) for more information.
  • International Conference on Survey Nonresponse, October 28-31, 1999, Portland Hilton, Portland, Oregon.

Table of Contents

To subscribe to this free newsletter send an e-mail to TMIP@tamu.edu or contact Gary Thomas at (ph.) (979) 458-3263, (fax) (979) 845-6001, (mail) Gilchrist, Room 112, Texas Transportation Institute, Texas A&M University System, 3135 TAMU, College Station, TX 77843-3135