TMIPConnection
Issue 6, November 1996

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

Innovative Projects Report — Maryland Department of Transportation

There is a widely held view that much innovative planning and modeling work is being done in states, MPOs, and local governments across the country, but that information on these efforts is not being distributed. At the recent Urban Design, Telecommuting and Travel Behavior conference this concern was raised in several breakout sessions and in the final plenary session. Howard Simons of the Maryland Department of Transportation (MDOT) offered the Transportation Emission Reduction Pilot (TERP) Program as an example of work being done at the local level. With your help, future volumes of the TMIP newsletter will contain other articles describing innovative projects.

TERP is an innovative program to fund, implement, monitor and evaluate emission control strategies. The cooperative effort between MDOT and local jurisdictions was proposed in 1994 to address regional air quality conformity in the Baltimore and Washington nonattainment areas at a time when State and local governments were attempting to develop transportation related emission reduction strategies. There was concern that emission reduction measures used in other areas of the country would not be successful in Maryland. Transportation trust fund grants were made available to local jurisdictions in the Baltimore and Washington regions to implement emission reduction projects. Funding for TERP was programmed at $3 million for FY 96 and $2 million for FY 97. Local jurisdictions provided a 20% match and did a before and after study to determine if the emission reduction project was successful and cost effective.

Two TERP Advisory Committees, one from Washington, the other from Baltimore, were formed which included officials from the transportation planning offices of the local jurisdictions, from MDOT, and the Maryland Department of Environment. The Committees, which met separately, reviewed and approved program criteria and identified a variety of projects in their respective regions. The proposals submitted by the local jurisdictions were reviewed and massaged by the committees and then forwarded to the TERP Executive Committee for the final decision on grant approvals.

Grants for more than 40 emission reduction projects requested by 14 local jurisdictions were awarded in August 1995. Each jurisdiction was requested to sign a grant agreement which required the jurisdiction to submit an annual report and a detailed description of a before and after study to evaluate the emission reduction benefit for each of its projects.

The types of projects which were approved for grants and which are now in various stages of planning/implementation are:

  1. vanpool incentives;
  2. bicycle/pedestrian pathway enhancements;
  3. bus stop enhancements;
  4. transit marketing programs and subsidies;
  5. alternative fuel vehicle conversions and refueling stations;
  6. guaranteed ride home programs;
  7. shuttle service between public transit and major workcenters, and;
  8. telecommuting.

The results of these programs will be reported in future newsletters.

This description of the TERP Program was reprinted from an article appearing in the July, 1996 issue of Maryland on the Move, a publication of the Maryland Department of Transportation, provided by Gil Weidenfeld and Howard Simons. If you have innovative projects which you think would be of interest to others in the planning profession, please contact Kim Fisher at (202) 366-4054.

Table of Contents

Travel Demand Forecasting Seminar

Seminar Concept

In its role as technical assistance and outreach arm of TMIP, TTI is going to offer a series of technical exchange seminars around the country. The seminars are open to travel demand forecasting modelers. Attendance will only be limited by participants' willingness to travel, although we hope to minimize travel costs by offering the seminars in several locations. The one-day seminars will focus on specific travel demand forecasting topics of interest to local/regional travel demand modelers. In order to ensure the relevance of the topics, we will look to the practitioners working in each area to help define the topic.

The seminars will open with a brief presentation by an expert in the chosen forecasting topic (experts may be TTI staff, a consultant, Department of Transportation staff person, or a modeler from the region). The remainder of the day will be an open discussion of the chosen topic and other areas of concern. The first seminars will be open to both intermediate and experienced travel demand forecasting modelers. We are considering creating two seminar sessions, one for intermediate modelers and another for experienced modelers. If participation is high and it appears that the discussions are more productive in the two-level format, we may move to adopt that structure in the future.

Seminar Goals

The seminars have several goals — the most obvious of which is to disseminate new techniques. The introductory presentation in each seminar will describe the state-of-practice and the state-of-art in the topic areas and briefly explain how modelers around the country are addressing problems. This will provide both modeling ideas and potential contacts for those seeking more information.

A less obvious goal will be to help establish communities of modelers in each region. Increasing work load combined with shrinking financial resources and staff has resulted in less contact and interchange among travel demand forecasting professionals. Our sense is that there are many creative solutions, modeling techniques, and research efforts underway at the local and state level which would be very valuable if they were more widely known. We hope that these seminars, combined with other parts of the TTI outreach effort, will improve dissemination. TMIP staff, including the federal sponsors, will have the additional benefit of meeting with modelers at the local level. As a result, this will increase the TMIP staff's knowledge of the problems facing those modelers.

Seminar Topics

The topics for each seminar will be determined by the needs of the modelers in that particular area. However, based on technical assistance we have provided through different mechanisms, we have already identified several potential topics:

  • Windowing or subarea analysis
  • Incorporating urban form in traditional modeling
  • Implementing feedback loops
  • 4-step models
  • Forecasting VMT for conformity
  • Bike and pedestrian modeling analysis
  • Allocation of regional land use forecasts to TAZs

Next Step

Beginning next year, TTI will contact potential hosts and regions for the upcoming seminars. We welcome suggestions on either seminar locations or topics of interest to you. Gordon Shunk, at (817) 277-5503, or Kim Fisher, at (202) 366-4054, would be happy to speak with you on this or other technical assistance requests.

Table of Contents

TMIP Review Panel

The TMIP Review Panel was established to provide comments on the design and implementation of program activities. The last TMIP newsletter introduced the two newest TMIP Review Panel members, Annette Liebe and Keith Killough. After publishing that newsletter, however, we realized that the existing, and long-suffering, panel members had not been properly presented. To right this wrong, we will begin to highlight two panel members in each newsletter.

Martin Wachs, the current chairman of the Review Panel, recently became the Director of the University of California Transportation Center at the University of California, Berkeley. Prior to July 1996, he was Professor of Urban Planning and Director of the Institute of Transportation Studies at UCLA, where he had been a member of the faculty since 1971, and where he served three terms as Head of the Urban Planning Program. Professor Wachs holds a Bachelors Degree in Civil Engineering from the City University of New York, and M.S. and Ph.D. degrees in Transportation Planning from the Civil Engineering Department at Northwestern University and University of Illinois at Chicago.

Dr. Wachs is the author/editor of four books and has had published over one hundred articles on transportation planning and policy. His most recent book written with Mark Garrett, Transportation Planning on Trial — The Clean Air Act and Travel Forecasting, was published in 1996. Dr. Wachs has received numerous awards and fellowships, has served as an editor for many journals, and has been a transportation consultant to many private and non-profit institutions. His breadth of knowledge and experience have been invaluable to the TMIP program.

Michael Morris has been on the staff of the Transportation Department of the North Central Texas Council of Governments (NCTCOG), the Metropolitan Planning Organization for the Dallas-Fort Worth area, since 1979. He became Deputy Director of the Department in 1983 and the Director in 1990. Mr. Morris is responsible for the overall activities of the Transportation Department including the implementation of the Regional Transportation Plan, Transportation Improvement Program, Congestion Management System, and air quality-related Transportation Control Measures of the State Implementation Plan. He received his Masters in Civil Engineering from the State University of New York in 1979 and is a registered Professional Engineer in the State of Texas. He received the 1994 Transportation Engineer of the Year Award from the Institute of Transportation Engineers, Texas Section, and the Texas Department of Transportation Road Hand Award in 1995. He is a member of the Association of Metropolitan Planning Organizations Board of Directors.

Beyond the contributions Mr. Morris has made to the overall TMIP program, he and other NCTCOG staff have been instrumental in the design and implementation of the Dallas Interim Operational Capability (IOC) and Case Study (see the TRANSIMS TRAVELOGUE for a description of the IOC and Case Study.) NCTCOG's participation in this effort has been instrumental in defining the Case Study, identifying data collection techniques and problems, and by adding a practitioner's perspective to TRANSIMS development.

Table of Contents

Upcoming Conferences Related to TMIP

  • Third Travel Model Improvement Program Conference, December 8-11, 1996, San Diego, CA, Holiday Inn on the Bay, (800) 877-8920. Contact Lynette Engelke, (817) 277-5503 ext. 226, for further information.
  • Transportation Research Board, 76th Annual Meeting, January 12-16, 1997, Washington DC, Omni Shoreham/Sheraton Hotels. Contact the Meetings section of TRB, (202) 334-2362, for further information.
  • South West Transit Association 17th Annual Conference, February 2-4, 1997, Dallas, TX, La Meridien Hotel. Contact Carol Ketcherside, (210) 340-8726, for further information.
  • Transportation Research Board, 6th Conference on Transportation Planning Methods and Applications, May 19-23, 1997, Dearborn, MI. Contact Paul Hershkkowitz, (517) 373-9038, for further information.
  • ITS America 7th Annual Meeting, June 2-5, 1997, Washington, DC, Sheraton Washington. Contact Bonnie Jessup, (202) 484-2896, for further information.
  • Institute of Transportation Engineers 1997 Annual Meeting, August 3-6, 1997, Boston, MA. Contact ITE, (202) 554-8050, for further information.

Table of Contents

TMIP and ITS

by Fred Ducca and John Harding

Intelligent Transportation Systems (ITS) pose a unique challenge to the travel forecasting process. How does a process, designed to estimate the impacts of major traditional improvements such as new highways or rail systems, reflect the influence of the application of technology (adaptive traffic signal control systems, incident detection) to the transportation system? Modelers can easily add new links to a highway network or add a rail line to a transit network, but currently they cannot look at systems that respond in real-time to network conditions indicated by an array of sensors.

Many areas now have aspects of ITS in early deployment and plan additional deployment. The DOT has the objective, through Operation TimeSaver, of deploying ITS in 75 urban areas within ten years. In response to this ITS activity, and the need to update travel models, the TMIP is cooperating with the ITS joint Program Office and with the Office of Safety and Traffic Operations Research and Development to remedy the deficiencies in ITS analysis and evaluation capabilities. TMIP, the Joint Program Office, and the Office of Safety and Traffic Operations Research and Development will jointly develop a near term Track B type solution to meet immediate needs and in the long term will integrate ITS analysis capabilities into TRANSIMS.

The Intelligent Transportation Infrastructure Deployment Analysis System (IDAS) project encompasses the efforts to provide a sketch planning analysis capability to support near term ITS deployment analysis and evaluation. IDAS builds from the basic four step modeling process by providing capabilities to enhance networks to represent ITS, analyze the impacts of ITS, provide life cycle cost estimates, and compare the results of alternative improvements. The figure below represents the conceptional framework for IDAS:

IDAS Framework

The following describes the various IDAS components.

Alternative Network Generator

Purpose: Supports disaggregation of data to a detailed micosimulation level, and the representation of ITS supporting infrastructure. Provides functions that enable, through a network depiction of the transportation system, the indication of ITS and sequencing of improvements (conventional, ITS, and combinations) over a 20 year planning horizon.

Input: Existing network data; planning networks, GIS data, information on network characteristics.

Output: Alternative networks that contain information on improvement attributes including ITS. Provides the information necessary to conduct ITS cost analysis and the structure and information needed for ITS impact analysis.

Enhanced Impact Analysis

Purpose: To facilitate the analysis of ITS impacts using existing network, performance, and best knowledge concerning the nature of ITS impacts. This module calculates impacts due to ITI improvements on an individual and system basis over a 20 year planning horizon.

Input: Alternative network files from the Network Generator module, and transportation analysis outputs from conventional (four step) travel forecasting procedures.

Output: Amended parameters that indicate the impacts of incorporating or introducing ITS improvements. Various amended parameters may service other analyses such as Air Quality and Fuel Consumption. All parameters are used to generate files and Measures of Effectiveness that support presentation of impact results, and alternatives comparison analysis.

Infrastructure Cost Analysis

Purpose: To calculate and distribute the life cycle costs of the supporting infrastructure necessary to operate and integrate the various ITS improvements in each alternative. The cost calculations will account for all aspects of ITI broken-down into project phases over a 20 year horizon.

Input: Alternative network files from the Alternatives Generator module. Cost figures, parameters for basic conventional infrastructure components, and cost factors for the 20 year horizon.

Output: Cost figures for each alternative and specific improvements. Output will consist of raw cost data for the development of various presentation graphics, and a cost file formatted for use with the Alternatives Comparison module.

Alternative Comparison Analysis

Purpose: Supports comparison of alternative improvements through a generalized multi-level criteria structure. The module will provide information concerning the merits of various alternatives and improvements.

Input: Alternative network files, impact analysis parameters and MOEs, criteria weighting.

Output: Alternative and individual improvement ranking indicators and criteria sensitivities.

The overall IDAS structure and time frame have not been fully specified and await further decisions on available funding.

ITS and TRANSIMS

In the long term, TRANSIMS will develop the capability of representing ITS. During FY97, plans call for specifying ITS functions and TRANSIMS representation of these ITS functions. TRANSIMS, for example, will represent system surveillance and detection capabilities, adaptive traffic signal control, and the system functions needed to enable the communications between them. Through this effort, an array of ITS functional representations will be incorporated into TRANSIMS that will enable the evaluation of the various ITS components, e.g., Advanced Traffic Management Systems, Advanced Traveler Information Systems. The incorporation of ITS functions will be accomplished through a case study procedure that will culminate in testing these TRANSIMS capabilities in a field application.

The functional representations present an interesting conundrum to modeling; often times the easiest functions to implement are the hardest to model, and vice versa. Variable message signs, as an example, can quickly and cheaply be installed, but driver response is difficult to predict. Adaptive Traffic Signal Control Systems are costly and complex to implement, but the impacts on travel are far easier to depict due to the regulatory nature of traffic signals. Assessing the impact of a series of region-wide micro-level changes on overall travel presents a complex challenge which TRANSIMS developers look forward to meeting.

Table of Contents

New Report Notices

Several new reports have been published since the last newsletter, both by TMIP and others, which we think would be of interest to the TMIP audience. The following descriptions include ordering information — please note that because these are not all TMIP products, you will need to go directly to the source to obtain the documents.

Travel Survey Manual

The Department of Transportation, in conjunction with the Environmental Protection Agency, has just released a new Travel Survey Manual. It has been 23 years since federal travel survey guidance was last updated. The first set of travel survey guidelines was published by the Bureau of Public Roads in the 1940s, updated in the mid-1950s and again in 1973. The newly published Travel Survey Manual incorporates the many changes in the survey, planning, marketing, and computer fields which have occurred since the earlier manuals were published. Examples of these changes, which are covered in the manual, include declining participation rates, increased modeling data requirements, and technology advances in surveying such as computer assisted telephone interviewing (CATI). This manual is intended to help individuals responsible for implementing travel surveys to avoid some common pitfalls and to ask intelligent questions when working with survey consultants.

This manual opens with an overview of the survey process and discussions of issues common to most survey manuals: management and quality control, sampling, and precision and accuracy in surveys. The overview is followed by seven chapters covering the most common transportation surveys, including household surveys, transit on-board surveys and others. The final two chapters explore the emerging use of stated-response surveys and longitudinal surveys and techniques for geocoding data. The appendices include information on the costs of surveys and examples of Requests for Proposals used in recent contracting processes. Field manuals and interviewer manuals are also included.

The manual can be ordered from the US DOT at: TASC Subsequent Distribution Office, Ardmore East Business Center, 3341 Q 75th Avenue, Landover, MD 20785, Fax (301) 386-5394, E-mail: SDS.Info@ost.dot.gov. When ordering, the complete title, Travel Survey Manual and Appendices, and the publication number, FHWA-PL-96-029, must be included.

Characteristics of Urban Freight

This report has been developed to support the transportation planning needs for urban goods movement and freight planning. This is a compilation of current data pertaining to urban freight movements. The data was assembled from many different sources and is expected to be of assistance to Metropolitan Planning Organization planners who handle urban freight issues. The information, drawn from U.S. and Canadian experience, is community specific as it is not yet possible to develop generalized relationships. As more data becomes available from current data collection efforts, it should be possible to develop generalized values which can be transferred between planning environments. All data was obtained from survey studies and was not synthesized from analytical modeling efforts. All data sources have been identified to assist the planner. Most of the information has been collected from published reports, but some data, particularly in the intermodal freight area, came from internal memos, personal observations, and interviews. To receive a copy of this document please contact: Stefan Natzke at Federal Highway Administration, Intermodal & Statewide Program Division, 400 7th Street SW, HEP-10, Washington, DC 20590, or call (202) 366-9236.

Environment and Planning Resource Catalog

FHWA's Office of Environment and Planning (OEP), working with its counterparts in the Federal Transit Administration, has recently completed production of a Planning and Environmental Resources Catalog. Jointly prepared as a service to our planning and environment customers, the catalog is a guide to the vast array of resources, both human and informational, which we have developed in support of transportation planning and environment enhancement. The catalog is one part of a multifaceted strategy to ease the difficulty our customers have in obtaining assistance on technical issues, as well as matters of policy and guidance.

Included in the catalog are the many products and services developed by FHWA and FTA since ISTEA and the Clean Air Act Amendments significantly changed the practices of transportation planning and environment analysis. Specifically, customers will find information on the availability of printed reports, videos, course handbooks, training, "on-line" databases and statistics, conference proceedings, and federal personnel (including how to contact individual staff). The information is sorted by air quality; environmental, social and resource impacts; transportation finance; general planning; statewide planning; and metropolitan planning.

The manual can be ordered from the US DOT at: TASC Subsequent Distribution Office, Ardmore East Business Center, 3341 Q 75th Avenue, Landover, MD 20785, Fax (301) 386-5394, E-mail: SDS.Info@ost.dot.gov. When ordering, the complete title Planning and Environmental Resources Catalog and the publication number, FHWA-PD-96-039, must be included.

The previously published Environment and Planning Training Catalog may also interest transportation professionals as it describes the planning and environmental training courses, workshop, and seminars offered by the Federal Highway Administration and the Federal Transit Administration. To receive this catalog please contact: Brian Jackson, Federal Transit Administration, 400 Seventh Street, SW, TPL-20, Room 6100, Washington, DC 20590, (202) 366-2360.

Table of Contents

Myths and Facts of TRANSIMS

TMIP staff, federal sponsors, and the Los Alamos National Lab (LANL) TRANSIMS staff spend a lot of time trying to answer what we have come to regard as the "TRANSIMS myths". The following is a somewhat tongue-in-cheek takeoff of the Myths and Facts of Transportation which Bob Dunphy of the Urban Land Institute has produced. This is only somewhat tongue-in-cheek because the facts are facts — most of the time!

Myth #1: TRANSIMS brings down the electric grid when turned on. Remember the brown out in the western United States last summer — the Dallas IOC!
FACT: As far as we know no power providers have tried to blame TRANSIMS for their problems, but they would surely like the idea.

Myth #2: Our grandchildren will use the TRANSIMS proto-type — maybe.
FACT: The addition of ITS and lower than expected funding has slowed the progress on TRANSIMS. However, the current schedule calls for completing TRANSIMS in the year 2001.

Myth #3: TRANSIMS requires ten Cray computers and a blood sacrifice to run.
FACT: TRANSIMS is currently running on networked ( a local area network or LAN) SUN stations. This type of computer is becoming increasingly common in transportation planning agencies, particularly those using GIS. It is anticipated that when TRANSIMS is available, in 2000, that the price and analytical capabilities of desk top computers will make the hardware question insignificant. The blood sacrifice is optional, but we suspect it will work without one.

Myth #4: TRANSIMS will take one week to simulate a 24 hour period of travel.
FACT: TRANSIMS can run simulations faster than real-time. For example, in the Dallas IOC, using 4 machines, TRANSIMS simulated five hours of travel with 280,000 vehicles in less than five hours.

Myth #5: TRANSIMS can be run without any data — it uses black magic.
FACT: See Myth #9.

Myth #6: Simulations are better than virtual reality and/or real reality.
FACT: This is a myth that LANL may be guilty of propagating. Dick Beckman is so persuasive when describing his population synthesizing techniques!

Myth #7: TRANSIMS is endorsed by the cast of Star Trek.
FACT: Actually only the Deep Space Nine cast has endorsed TRANSIMS. Casts of the other shows are withholding judgement until they see the reports on the Dallas IOC.

Myth #8: Fred Ducca is planning on building a media career from the TRANSIMS project.
FACT: Fred is going to keep his day job, but his picture did appear in the Philadelphia Inquirer with a TRANSIMS article.

Myth #9: TRANSIMS data requirements include coding every single road in the region with all pothole locations.
FACT: The data requirements for TRANSIMS will be greater than for a traditional 4-step model -- but less than would be required for traffic operation models. The challenge facing the TRANSIMS staff and others interested in the project is that, unlike the traditional applications of operational models which simulate one corridor at a time, TRANSIMS will simulate an entire region. This does mean that the data requirements could be extensive. For this reason testing different levels of data aggregation and disaggregation plays an important role in the TRANSIMS development effort. Currently, for example, the TRANSIMS staff and the NCTCOG staff are testing the sensitivity of the microsimulation to the use of a generic collector street network versus the realistic coding of all collector streets. The goal is to develop TRANSIMS with the simplest data requirements and still allow the evaluation of air quality and the other difficult questions TRANSIMS is designed to address.

Myth #10: TRANSIMS will run on a 386 ... really. With Windows.
FACT: While many of us with old machines would like to believe they have a very long functional life this is probably unrealistic. See Myth #3.

Table of Contents

Listing of Available Reports

  • Activity-Based Modeling System for Travel Demand Forecasting, Final Report, September 1995, Report DOT-T-96-02. Contact: Subsequent Distribution Center
  • The Effects of Added Transportation Capacity: Conference Proceedings, Bethesda, MD, Report DOT-T-94-12. Contact: Subsequent Distribution Center
  • The Effects of Land Use and Travel Demand Management Strategies on Commuting Behavior, July 1994, Report DOT-T-95-06. Contact: Subsequent Distribution Center
  • Identification of Transportation Planning Data Requirements in Federal Legislation, July 1994, Report DOT-T-94-21. Contact: Subsequent Distribution Center
  • Incorporating Feedback in Travel Forecasting: Methods, Pitfalls, and Common Concerns, March 1996, DOT-T-96-14. Contact: Subsequent Distribution Center
  • Multi-Criteria Equilibrium Traffic Assignment Basic Theory and Elementary Algorithms: Part I, T2: The Bicriteria Model, August, 1994. Contact: Kim Fisher
  • Network-Optimized Congestion Pricing: A Parable, Model and Algorithm, May 1995. Contact: Kim Fisher
  • An Operational Description of TRANSIMS, June 1995. Contact: Kim Fisher. Peer Review Panel Functions and Organization. Contact: Kim Fisher
  • Planning and Environmental Resources Catalog, November 1996, FHWA-PD-039. Contact: Subsequent Distribution Center
  • Short-Term Travel Model Improvements, October 1994, DOT-T-95-05. Contact: Subsequent Distribution Center
  • Summary of Comments Prepared by Travel Forecasting Peer Review Panels. Contact: Kim Fisher
  • TRANSIMS Model Design Criteria as Derived from Federal Legislation, June 1995, Report DOT-T-95-21. Contact: Subsequent Distribution Center
  • TRANSIMS Project Description: Travel Model Improvement Program, August 1994. Contact: Kim Fisher
  • TRANSIMS: TRansportation ANalysis and SIMulation System: Project Summary and Status, May 1995. Contact: Kim Fisher
  • Travel Model Improvement Program: Conference Proceedings, August 14-17, 1994, Report DOT-T-95-13. Contact: Subsequent Distribution Center
  • Travel Model Improvement Program: Land Use Conference Proceedings, February 19-21, 1995, December 1995. Contact: Kim Fisher
  • Travel Model Improvement Program: Project Summaries, September 1995. Contact: Kim Fisher
  • Travel Survey Manual, November 1996, FHWA-PL-96-030. Contact: Subsequent Distribution Center

Contact information for report contacts:

Kim Fisher
c/o Federal Highway Administration
400 7th St., SW
HEP-22, Room 3232
Washington, DC 20590
Phone (202) 366-4054
Fax (202) 366-3713
E-mail: kim.fisher@fhwa.dot.gov

Subsequent Distribution Center
U.S. Department of Transportation
Ardmore East Business Center
3341 Q 75th Avenue
Landover, MD 20785

Table of Contents

TRANSIMS Travelogue — November 1996

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 the TRANSIMS 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 interim operational capabilities (IOC) to cover the major TRANSIMS components: Household and Commercial Activity Disaggregation, Intermodal Route Planner, Transportation Microsimulation, and Environment (primarily air quality). As each IOC is ready and with the collaboration of a selected 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.

The Traffic Microsimulation is emphasized in the first IOC, which we are testing currently. We are working with the selected MPO, North Central Texas Council of Governments (Dallas-Fort Worth), on the case study that the IOC should support.

Cellular Automata Microsimulation

In a previous Travelogue we discussed very generally the cellular automata (CA) approach to traffic microsimulation. In this Travelogue we present additional detail about the CA methods developed for the current TRANSIMS IOC and applied in the Dallas-Fort Worth case study. The following discussion describes the fundamental model, the emergent traffic dynamics, its theoretical basis, possible extensions, calibrations, and data smoothing for emissions calculations. The discussion is based primarily on the work of Kai Nagel and his collaborators documented in the references following this article. This CA model also is often called the particle hopping model.

Basic Model

The fundamental CA model considers a single-lane freeway. The freeway length is sectioned into an array of cells of uniform length. Each cell's length is the average distance (approximately 7.5 m) between vehicles when traffic is at a complete standstill, that is, in jammed traffic. A cell may be empty or contain a vehicle. If it contains a vehicle, the vehicle has an integer velocity between zero and a maximum velocity, Vmax = 5. The integer velocity represents the number of cells that the vehicle moves the next step. The step size is exactly one second, in which case Vmax corresponds to 135 km/hour, or about 84 mph. This step size abets fast computation because the updated vehicle position is computed by integer arithmetic and without multiplication of velocity and time step.

Updating the vehicle's next velocity and position is quite simple. First, we define the number of unoccupied cells ahead of the vehicle as its "gap." Then, we update the velocity by accelerating to the maximum velocity without running into the vehicle ahead:

V(t+1)=min[V(t)+1, Vmax, gap].

But, with probability P, we reduce this tentative velocity by one (without going backwards):

V(t+1) = max[V(t+1) - 1, 0].

Finally, we update the vehicle's position:

X(t+1) = X(t) + V(t+1).

This rule set is called the Nagel-Schreckenberg model. The random velocity reduction process captures driver behavior such as free-speed driving fluctuations, non-deterministic accelerations, and overreactions when braking. With a deceleration probability of 0.5, the average free speed is approximately 75 mph.

Emergent Traffic Dynamics

This simple model produces dynamics observable in everyday freeway traffic. First, we can display an individual vehicle's movement in space and time as shown in Figure 1. Vehicles moving at constant velocity leave straight-line tracks slanting downward to the right. A stopped vehicle moves in time, but not in space, creating a vertical line. The figure shows the spontaneous formation of well-known traffic shock waves that propagate backward in space.

figure shows the spontaneous formation of well-known traffic shock waves that propagate backward in space

This model also produces the fundamental flow-density relationship shown in Figure 2 where density has been normalized to 1.0 for a completely jammed, nonmoving system. A comparable plot can be generated from real traffic measurements. At low densities, flow increases linearly with more vehicles in the system. Near a density of 0.1 the system achieves maximum throughput or 'capacity,' but the flow is quite chaotic and its variability increases dramatically. In this density region the average travel time increases, and the travel time variance jumps tremendously. At higher densities traffic disturbances spread throughout the system until the system comes to a complete standstill at a density of 1.0.

fundamental flow-density relationship

CA model results illustrate that the most efficient state from a traffic flow perspective is at the transition between low-density free flow and high-density, long-lifetime traffic jams. But, in this state spontaneous small fluctuations can cause large emergent traffic jams. Furthermore, as seen in Figure 1 jams themselves cause branching jam waves commonly observed as stop and go traffic. Jam wave perseverance and repeated branching produce correlated jam waves even though, from a traveler's viewpoint, their relative spatial separation may indicate no apparent common cause.

Car-Following Models

We compared the cellular automata approach with car-following models for vehicular traffic. Car-following models typically consider following distances, time headways, driver reaction times, vehicle inertia, etc. The inherent one-second time step of the CA model implicitly represents driver/vehicle reaction time delays and minimum following times. Furthermore, a local vehicle control system, that is, an adaptive "driver" who reacts to his environment, emerges from the simple CA rules. The "driver" exhibits a breadth of responses (velocity adjustments), dependent on his current velocity and gap. Thus, this controller contains a higher fidelity representation than apparent in the simple rule set.

It is not intuitively obvious that the hopping behavior of an individual CA vehicle traversing a roadway network bears resemblance to reality. In one sense we are not concerned with the individual vehicle's behavior. We are concerned whether the behavior of the ensemble of CA vehicles produces properties we can compare with measured traffic dynamics and whether we can use these properties to derive information about the performance of the transportation system. The CA model serves this function and computes fast, necessary to compute traffic flow over vast metropolitan transportation networks. Furthermore the simple CA rule set has "driver" characteristics, and careful analysis shows that the CA model can be considered a model of a driving model.

Classic Fluid Dynamics Models

Simplifications of this CA model correspond to certain cases of the Lighthill-Whitham theory, used in traffic theory for over 40 years. For instance, a CA model with Vmax=1 and random movement of vehicles to unoccupied adjacent cells corresponds in the limit to Lighthill-Whitham theory with added noise and diffusion and specialized to the Greenshields flow-density relation where flow = (1-). If we leave Vmax=5 and remove the random component, the model corresponds to the fluid-dynamics continuity equation with a wave velocity of Vmax in light traffic and -1 in heavy traffic. This is Lighthill-Whitham theory with another flow-density relation but without noise or diffusion. Thus one can show that certain aspects of the CA model traffic jam dynamics are phenomenologically the same as in fluid-dynamics traffic models. Yet, the CA model includes fluctuations, which fluid dynamics theories do not.

Extensions

The basic Nagel-Schreckenberg CA rule set does not produce the close-following behavior usually observed in high-speed traffic. As a result, the maximum capacity displayed by the model is somewhat lower than measured on single and multiple lane roads. This model disparity can be overcome by redefining the current vehicle's gap to account for the next vehicle's velocity and gap.

The coarseness of the basic CA grid and update rules can cause concerns about spatial resolution. For example, a significant portion of the traffic may comprise vehicles larger than the cell size, or the single step accelerations may be excessive for emissions modeling, or the speed variability may be excessive for local street speed limits (Vmax). Finer spatial resolution can be obtained with further grid subdivision, but must carefully maintain the model implicit features such as reaction times, following times, and jam spacings. The rules themselves must account for vehicle lengths and jam densities whereas previously the vehicle lengths and jam densities were implicit in the grid spacing and the rules.

A major extension is to model traffic on multiple lane roads in which vehicles change lanes and pass other vehicles. Now the rules consider not only the vehicle's gap ahead, but also the adjacent lane cell occupancies in both directions (to avoid sudden stops and rear-end collisions). Various rule sets can be devised, but a simple one is to move to an unoccupied adjacent cell in the adjacent lane if:

  • V(t+1) > gap ahead in current lane, and
  • V(t+1) < gap ahead in adjacent lane, and
  • Vmax < gap behind in adjacent lane.

To keep vehicle platoons from bouncing between lanes, we add an additional requirement that lane changing also occurs randomly with some probability. After the lane change, the Nagel-Schreckenberg rules are applied as before.

A consideration for the multiple lane rules is whether to bias the traffic to use the right lane in the absence of other traffic constraints. Some CA rule implementations may inherently contain a lane bias in which the results depend on the update order for lanes or vehicles. If such lane asymmetries are undesirable or don't match lane usage or lane change measurements, the update order may be chosen randomly with probabilities that yield the desired behavior.

Not all motorized travel occurs on a freeway. A TRANSIMS street intersection is designed to capture the associated time delay, not the detailed turning dynamics or the intersection geometry other than possible turn bays and merge lanes. The intersection model contains allowed movements from incoming to outgoing lanes. Signalized intersections have timing and phasing plans with protected and unprotected movements. Unsignalized intersections may have stop, yield, or no signs. CA vehicles within the intersection enter buffers that capture the delay associated with passing through the intersection. Bufferless intersections assure that a freeway-ramp intersection does not perturb the CA freeway dynamics in the absence of ramp traffic. Entry onto a new roadway segment (link) requires a gap in the targeted cells, and unprotected entry also requires a gap in the traffic competing for the same road and a gap in cross traffic.

We are not developing TRANSIMS just to replicate traffic dynamics, but to examine the relationships and interactions between the transportation system and the traveling population. To understand how the transportation system affects individuals and the decisions they make about traveling, we must follow each individual's travel. Thus, each traveler has a trip plan defining his planned departure time and detailed route and transportation modes. For this IOC the trip plan is assigned to the CA vehicle. Whenever the CA vehicle enters a new link, the trip plan's next link and the next intersection's allowed movements establish the "plan lane(s)" in which the vehicle must be to stay on the plan. Thus, in addition to the basic lane changing rules, being in the proper "plan lane" is another reason for changing lanes. As the vehicle nears the intersection, the "plan lane" rule gradually overrides the other rules that may otherwise inhibit lane changing (except if the adjacent cell is currently occupied).

Calibrations

Before the CA microsimulation can execute an arbitrary demand on a complex roadway network, it should perform properly in simple controlled situations. Then we would be assured that the CA microsimulation would not cause unrealistic or confusing results in complex situations. So, we designed several simple traffic experiments with controlled demand to calibrate the CA microsimulation. As shown in Figures 1 and 2 a single-lane circle with various traffic densities calibrates the car-following behavior. We use two- and three-lane circles to calibrate lane changing behavior and lane usage as well as to establish the flow-density relation for multi-lane traffic. In another test, vehicles with specified plans through an intersection randomly begin on one of three lanes (left, through, right) heading toward the intersection to verify the model's plan-following capability.

We designed an intersection with traffic merging onto a major highway, measured the merging-traffic volume as a function of the known through-traffic volumes, and obtained results comparable to Highway Capacity Manual data. Similarly, for left turns against oncoming traffic, we measured turn volumes against known oncoming traffic volumes and obtained results comparable to Highway Capacity Manual data as well as to results from another microsimulation method. We also measured the headway distribution of vehicles leaving an intersection and obtained results similar to actual traffic measurements. Such calibration experiments are the start of a suite of tests to verify the CA microsimulation performance for a variety of situations.

Emission Implications

Direct observance of the CA vehicle hopping motion gives quantum velocities and accelerations. These velocities and accelerations are unrealistic for emissions model input. We are developing an approach using a Kalman filter to produce realistic, smoothed vehicle trajectories for the emissions module. The Kalman filter is designed for a physical process that has random elements and is observed with a noisy measuring device. We developed a formulation in which we estimate the fraction of aggressive drivers and the degree of their aggressiveness as expressed by their desired accelerations at a given speed.

We tested the formulation against arterial and freeway segments drawn from the California Air Resources Board data. The model addresses diverse driving situations, although further refinement may be useful. The situations we have modeled so far include arterial traffic with some vehicles starting from a stop and others continuing through a traffic light, uncongested freeway traffic and very congested freeway traffic.

In one idealized circumstance we used data from three cities to define conditional probabilities of accelerations among vehicles that had started from a stop and consistently accelerated to freeway speeds. We wanted acceleration patterns appropriate to uncongested conditions. However the procedure selected a very aggressive subset of the drivers.

In the test we used the actual speeds and accelerations as the desired driving behavior and mapped the trajectories into 7.5-m CA cells with corresponding speed intervals. Using the Kalman filter, we calculated smoothed trajectories and compared them to both the real trajectories and the CA trajectories for the same aggressiveness. The Kalman filter produced very accurate representations of the speeds and emissions from vehicles under these ideal conditions.

Summary

The cellular automata approach to traffic microsimulation produces traffic dynamics typical of that observed on freeways. It is computationally fast so that major metropolitan region traffic can be simulated in reasonable times. It has elements of car-following and fluid-dynamics traffic models, but has advantages over both. It has been extended to multiple lanes, city street driving, and trip plan following. Potential limitations of the basic model can be overcome by additional extensions. We have a test suite to calibrate the model against desired behavior and real data. Techniques exist to smooth the raw CA output for emissions modeling. In conclusion, the CA model has the attributes required for the TRANSIMS traffic microsimulation.

References

K. Nagel and S. Rasmussen, "Traffic at the Edge of Chaos," Artificial Life IV: Proceedings of the Fourth International Workshop on the Synthesis and Simulation of Living Systems, R. Brooks and P. Maes, eds., MIT Press, Cambridge, MA, 1994.

K. Nagel, "Particle Hopping Models and Traffic Flow Theory," Physical Review E, Vol. 53, No. 5, p. 4655, 1996.

C. Barrett, M. Wolinsky, and M. Olesen, "Emergent Control Properties in Particle Hopping Traffic Simulations," M. Paczuski and K. Nagel, "Self-Organized Criticality and 1/f Noise in Traffic," Traffic and Granular Flow, D. E. Wolf, M. Schreckenberg, and A. Bachem, eds., World Scientific, Singapore, 1996.

K. Nagel, M. Schreckenberg, M. Richert, and A. Latour, "Two-Lane Traffic Simulations Using Cellular Automata," Physica A (North-Holland).

C. Barrett, S. Eubank, K. Nagel, S. Rasmussen, J. Riordan, and M. Wolinsky, "Issues in the Representation of Traffic Using Multi-Resolution Cellular Automata," Los Alamos National Laboratory report LA-UR-95-2658.

M. Williams, G. Thayer, and M. Brown, "TRANSIMS Environmental Module," TRB A1F03 1996 Summer Conference, San Rafael, CA, 1996.

Further Information

For further information about the TRANSIMS program, please contact:

Dr. LaRon L. Smith
Los Alamos National Laboratory
PO 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

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