Click for next page ( 13


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 12
12 CHAPTER 4 THE TRAVEL BEHAVIOR RESPONSE MODULE The Portland Tour-Based Model was selected as the basis 1996. The model has several features that distinguish it from for the Travel Behavior Response Module because of its abil- traditional four-step travel models: ity to predict both modal and temporal shifts in travel behav- USER'S GUIDE ior as well as predict the impact on overall out-of-the-home Simultaneous modeling of trip generation, time of day, trip making. The Portland Tour-Based Model is complex, so mode choice, and destination choice. Utilities of lower- it is implemented in NCHRP Project 25-21 as a set of elas- level choices (e.g., mode and destination choice) are ticities rather than as the full model. incorporated in the utilities of higher-level choices (e.g., time of day and primary activity pattern). Application of the model to individual travelers. This 4.1 OVERVIEW OF THE PORTLAND approach, known as sample enumeration when applied TOUR-BASED MODEL to travel survey data, and more generally as microsim- ulation, is considered to be at the forefront of the cur- The Portland Tour-Based Model was originally developed rent state of the art in travel modeling. Microsimulation as part of a project to analyze road pricing policy alternatives allows the incorporation of detailed household and per- in Portland. An overview of the Portland model in a larger son characteristics that can significantly affect travel context is shown in Figure 2; the tour-based model proper behavior, such as presence of children in the household consists of the blocks within the large rectangle. (A full and competition for available cars in the household for description of the Portland Tour-Based Model is given in different trip purposes. Mark Bradley Research and Consulting, A System of Activity- Use of a synthetic sample to develop the base population Based Models for Portland, Oregon, Washington, D.C.: to which the model is applied. This approach provides Travel Model Improvement Program, U.S. Dept. of Trans- the model with a sufficiently large population so that com- portation, Report No.: DOT-T-99-02, U.S. Environmental plete trip tables can be produced. Sample enumeration Protection Agency, 1998. Consult this reference for details approaches based only on travel surveys generally pro- on model structure and coefficients.) duce results at a much larger scale, such as superdistrict- A more detailed look at the Portland model is given in to-superdistrict trip movements. The synthetic sampling Figure 3, which shows information flows between the dif- approach has been used for over 25 years. One early ferent submodels. The model system is designed to predict application was to the development of a database for the following: research on discrete-choice models. See Gerald Duguay, Woo Jung, and Daniel McFadden, "SYNSAM: A Meth- A full-day activity pattern (primary activity and, for tour odology for Synthesizing Household Transportation Sur- activities, subtour pattern), vey Data," Berkeley: Urban Travel Demand Forecasting Time of day (outbound, inbound) for home-based tours, Project, Working paper no. 7618, September 1976. Syn- Primary mode and destination, thetic sampling is currently used in the TRANSIMS Work-based subtours, and model and in the current version of the STEP model. An Location of intermediate stops. additional advantage of the synthetic sampling approach is that it enables disaggregation of benefit and cost esti- The Portland model is a conceptual descendant of Greig mates by socioeconomic category, which is often a sig- Harvey's Short-Range Transportation Evaluation Program nificant issue in transportation policy analysis. (STEP) model, with considerable additional detail. A descrip- tion of the STEP model and the theory behind the model is presented in Elizabeth Deakin and Greig Harvey's Trans- 4.2 DERIVATION OF ELASTICITIES portation Pricing Strategies for California: An Assessment of Congestion, Emissions, Energy and Equity Impacts: Final The Portland model has several drawbacks in application, Report, prepared for the California Air Resources Board, chief of which is the length of time required to operate it on

OCR for page 12
13 mp Input Tij ) with respect to travel time origin i to destination mp Employment by sector by TAZ j by mode m in time period p (denoted by t ij ). Sample of households Modal LOS measures For m = m and p = p, there is an own elasticity; otherwise, the quantity is a (mode or time or mode/time) cross-elasticity. Household-based tour model The quantities with tildes represent trips and travel times Primary activity after some change, and the other quantities represent base Secondary tour choice case trips and travel times. Time-of-day choice The equation can be converted to a log-log linear model: Mode/destination choice T~ijmp ~ijmp t ln mp = mp mp ln mp Equation 11 Tij m p t ij Work-based subtour model USER'S GUIDE Therefore, the elasticities can be estimated by observing the quantities Tijmp, T m p ~ijmp, tij ~ijmp predicted by the Port- , and t Intermediate stop location model (car driver tours only) land model and running a set of regressions against these results. The approach to generating the necessary data points was straightforward: Decompose tours to trips 1. Define a set of i, j zone pairs to be sampled. These zone pairs were sampled to focus on the areas of interest. For Output -- OD trip matrices by: Mode example, given the case study area, the research team Time of day focused on movements from within King County to Income group Seattle, from Pierce County to Seattle, and from Sno- homish County to Seattle. Movements to and from Kit- sap County were ignored because the research team Network Model believes that the ferry network may not be adequately (trip assignment by mode and time period) represented to treat this movement alongside bus tran- TAZ = traffic analysis zone. sit as a transit mode. LOS = level of service. OD = origin-destination. 2. Pick a particular zone pair with home zone i and desti- nation zone j. Randomly generate a travel time change Figure 2. Portland Tour-Based Model flow chart. in the AM peak period for the auto mode, and run the model only for the population within zone i. Store the relative change in travel time and the relevant changes even a high-speed computer. Consequently, it was decided to in travel by mode and time period as a data point. use the Portland model to develop a set of elasticities for pre- 3. Repeat Step 2 for different values of change to the dicting small changes in traveler behavior in response to indi- travel time. vidual traffic-flow improvement projects. The model was 4. Repeat Steps 2 and 3 for different time periods. executed several times on a range of travel time saving alter- 5. Repeat Steps 24 for different modes. natives, and the results were used to fit a set of demand/time 6. Repeat Steps 25 for different i, j zone pairs. elasticities. These elasticities were then incorporated into the 7. Collect the data points and run regressions on the appro- NCHRP 25-21 methodology. priate variables. A constant elasticity demand model in the following form was fitted to the Portland model: The research team believes that the following simplifica- tions were reasonable: mp ~ijmp T t~ijmp m p Tijmp = m p Equation 10 For small travel time changes, the constant elasticity mp tij approximation is probably good enough. It can be regarded as a first-order approximation to the demand Where: function. mp mp = the elasticity of demand for travel from origin i to Capacity improvements are likely to affect the peak peri- destination j by mode m in time period p (denoted by ods only. Hence, the main mode shifts are likely to occur