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 paper on the topic.3 The following points were covered in her presentation. ⢠MORPC, in cooperation with the Licking County Area Transportation Study, contracted with PB Consult, Inc. for the development of a new travel model in 2001. The MORPC area covers all of Franklin, Delaware, and Licking counties and portions of Fairfield, Pickang, Madison, and Union counties. The area includes 1,805 internal and 72 external traffic assignment zones. The 2000 population was approximately 1.5 million. The population is forecast to increase to 2 million by 2030. ⢠The new model is a disaggregate tour- based model applied with microsimulation of each individual house- hold, person, or tour. The model was completed in 2004 and further refined in 2005. The model consists of nine separate linked models and other network processing steps. The linked models include population synthesis, automobile ownership, daily activity pattern, joint tour generation, individual nonmandatory tour generation, tour destination choice, time- of- day choice, tour mode choice, and stops and trip mode choice. The core models are applied in a disaggregate manner. ⢠The model is applied with the microsimulation of each household, person, or tour, mostly using Monte Carlo realization of each possibility, estimated by the mod- els and a random number series to determine which possi- bility is chosen for that record. There are three global feedback loops for consistency between highway travel times that are used as input to, and as forecast outputs of, the model. Cube is used as the main model application package and TP is used for network management, assignment, external and commercial vehicle models, and other processing. After generating the networks and initial skims in TP+ and creating input files for a specific sce- nario, the custom Java programs are executed to implement the tour- based microsimulation models. The microsimulation results are aggregated in a preassignment process or step, which also integrates the commercial and external models to produce standard TP trip tables for four time periods, which are morning, midday, afternoon, and night. Vehicles are assigned from the final trip table with a multiclass equilibrium assignment utilizing 21 vol- ume and delay functions by facility and area type for each of the four time periods. There are three classes of com- mercial vehicles, four highway modes, and five transit modes. Nonmotorized modes are not assigned. ⢠The MORPC travel forecasting model can cur- rently run on three operational systems. One system is at PB Consult, one system is at MORPC, and one system was purchased by the Central Ohio Transit Authority (COTA). A system is also being developed at the Ohio Department of Transportation (ODOT), but is not yet operational. The initial system at MORPC was installed in 2004 and included one server and three worker com- puters. A fourth worker computer was added in 2005. The worker computers are directly networked and are isolated from the general MORPC network to make them less susceptible to viruses. This system is running a 32-bit Windows operating system and Java. ⢠COTA purchased a system to support the North Corridor Transit Project Draft Environmental Impact Statement. This system includes one server and four worker stations all running 64-bit Windows and Java. The system being implemented at ODOT will include one server and eight worker stations. It will use a dis- tributive version of Cube Voyager. ⢠The running times for the MORPC travel forecast- ing model for 2000 and 2030 vary by computer system. The total running time for the core model for 2000 ranges from approximately 20 hours to 35 hours. The total run- ning time for the core model for 2030 ranges from approximately 26 hours to 48 hours. The 2000 model includes 1.5 synthetic individuals making 2 million tours and the 2030 model has 2 million synthetic individuals making 3 million tours. The COTA system, which uses 64-bit computing, provides the shortest running times. Experience indicates that running antivirus software imposes a 15% penalty on the run time and that Win- dows will only allocate a maximum of 50% of memory to any one application. Planned future upgrades include installing Cube on the COTA work stations and sending TP scripts to run in parallel on the work stations. PREPARING PARCEL- LEVEL INPUT DATA FOR THE ACTIVITY- BASED TRAVEL MODEL IN SACRAMENTO Bruce Griesenbeck Bruce Griesenbeck described the use of parcel- level data in the development of a new activity- based travel model for the Sacramento Area Council of Governments (SACOG). He described the reasons for the use of parcel- level data, the general structure of the new model, and the approaches for forecasting street pattern and transit accessibility data used in the model. The following points were covered in his presentation. ⢠A major focus at SACOG over the past few years has been on land use policy and land use planning. The policy board recently adopted a long- range land use vision for the region, called the Blueprint. The vision provides a basis for other land use planning activities in the region. It promotes ideas related to urban containment and devel- 57THE SECRET IS IN THE SEGUE 3 See Anderson, R. S., Z. Jiang, and C. Parasa. Hardware Requirements and Running Time for the Mid- Ohio Regional Planning Commission Travel Forecasting Model. Volume 2, pp. 181â184.
