Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
⢠A key design issue at the person- day level is the number of activity and tour purposes to be considered. The early versions of the Portland model and the San Francisco model include three activity purposes. The other models have at least seven activity purposes. These activity purposes typically include work, school, escort, shopping, meals, personal business, and social and recre- ational. Most models do not distinguish between differ- ent types of in- home activities, including work and nonwork activities. ⢠A key feature of most of the models is the ability to model a personâs daily activity pattern. This feature includes how many trips individuals make for different trip purposes. In addition, some models include the pres- ence of extra stops by purpose, the allocation of stops to particular tours, and the presence of work- based sub- tours. Only a few models include in- home activities. ⢠The least consistent aspect of the different models is how tour complexity and trip chaining are measured. Some models consider these elements at the upper level of the model and all of the tour and trip decisions cas- cade off of this level, which results in a lot of substitution at the top level. Other models predict tours at the upper level of the model. These models predict if the individual makes stops and the purpose of the stop. ⢠Four types of household linkages are included in some of the models. First, the main pattern type for each person in the household is explicitly linked. A second linkage is joint tours with household members traveling together, which are generated separately for individual tours. A third linkage allocates some activity purposesâ such as escorting or chauffeuring, shopping, and other maintenanceâ between household members. A fourth linkage uses escorting as a tour or stop to pick up or drop someone off with a different activity purpose. These link- ages add complexity to a model. The Columbus model is the only model that includes all four linkages. ⢠Time- of- day models include simultaneously pre- dicting the time that an individual arrives at work and the time that he or she leaves work. A question is how narrow the time period should be for these models. One to 2 hours seems appropriate. At the trip level, where an individual may make intermediate stops along the tour, there is a need to predict what time the individual leaves each stop. In this case, a smaller interval, typically 15 or 30 minutes, is needed. The benefit of these types of mod- els is that they provide consistent tours across the day. Most of the models have fully consistent time windows. Every time an individual is predicted making a tour or a trip, that time window is blocked out so that the indi- vidual cannot do anything else. Time- pressure variables may also be used. These variables may result in individ- uals squeezing activities into shorter time periods or reducing the length of a trip. Time and space constraints can also be introduced into models. ⢠A major advantage of microsimulation models is that the analysis is completed at the zone level. The zone level can be used for origin and destination matrices, skims such as zone- to- zone travel times, and travel level of service. A more detailed level of spatial aggregation can be used for transit access times, walk times, and pedestrian environment factors. ⢠The Sacramento model can predict travel demand at the parcel level. There are 700,000 parcels in the model. A building level or parcel level is also being con- sidered in the Denver model. A hybrid approach of being able to use both a zone level and a parcel level may be a logical approach. The model design is similar for these applications. ⢠The models also allow accessibility from the land use and travel system to affect every single decision in the model system, not just mode choice and destination choice, but trip chaining, making more tours, automo- bile ownership, and work location. A traditional log- sum approach, where the whole model system is one large nested decision structure, would not be able to accommodate these elements, because it would take a very long time to run the model. Determining good log- sum or accessibility measures that can be used at the upper- level tour generation day pattern models and car ownership models may be needed in some models. Fairly detailed log- sums can be used for work and school desti- nations. Precalculated zonal mode and destination choice log- sums by segment can be used for transit acces- sibility, automobile availability, and household income. NEW YORK METROPOLITAN TRANSPORTATION COUNCIL TOUR- BASED MODEL DEVELOPMENT Peter Vovsha and Kuo- Ann Chiao Kuo- Ann Chiao and Peter Vovsha described the devel- opment and use of the New York Metropolitan Trans- portation Council (NYMTC) model. They discussed the need for a new travel model, the key elements of the model, and applications of the model. Volume 2 includes a paper on this topic.2 The following points were cov- ered in their presentation. ⢠Work began on developing a new travel model in the New York region in the late 1980s. One of the limi- tations of the old model was that it stopped at the Hud- son River. Household travel surveys were conducted in 28 counties in three states as part of the model develop- ment process. The region is approximately 9,700 square 14 INNOVATIONS IN TRAVEL DEMAND MODELING, VOLUME 1 2 See Vovsha, P., and K.-A. Chiao. Development of New York Metropolitan Transportation Council Tour- Based Model. Volume 2, pp. 21â23.
