Case Study:

Portland, Oregon

Methodology

Travel Impacts

Portland Travel Demand Model

Portland Metro's regional travel demand model is among the most advanced trip-based travel models in the U.S. It includes features such as household characteristics modeling; auto ownership modeling; market segmentation of households for trip generation, distribution and mode choice; time-of-day modeling; and feedback from congested traffic assignment to trip distribution and mode choice.. Three peak time periods are used: 7:00 to 9:00 a.m., 2:00 to 3:00 p.m., and 4:00 to 6:00 p.m. The model is run using EMME/2 software. Another noteworthy feature of the model, directly relevant to this analysis, is the inclusion of a relatively sophisticated truck model.

Metro recently undertook a major commodity flow survey (Portland Metro, 1997) to serve as a basis for the truck model. The commodity flow survey utilized public and private data sources on freight flows, in conjunction with external classification counts, to develop tables of movements by mode, commodity type, and direction of flow. Flows were tracked for 16 commodity groups. Commodity-flow ends were distributed to traffic analysis zones (TAZ) based on employment by industry in combination with specific information on flows through ports and airports. Commodity flows were then converted into truck movements through a series of processing steps, and truck volumes were converted into passenger-car equivalents. (Information on the commodities associated with truck movements is retained through this process.) Classification count data were used to split 24-hour demand into the peak periods required in the Metro model. For additional model documentation, see Cambridge Systematics (1998).

To develop forecast year (2020) in addition to base year commodity flows, regional economic forecasts were combined with judgments on shipping trends.

Application of Model

The Metro travel demand model was run for a base case and for each of the alternatives defined by the project team to provide forecasts for the year 2020. The model area was a six-county area covering Portland and surrounding areas in northwest Oregon and southwest Washington. For each alternative, Metro prepared peak passenger-car equivalent trip tables and ran three peak assignments of this table to the regional highway network.

The EMME/2 modeling software permitted Metro to segment the output trip tables into a maximum of 12 categories. The 16 commodity groups available in the Metro model were aggregated into eight categories of commodities moved by heavy truck. With the remaining four categories available, two categories of medium truck and two categories of auto, SOV and HOV, were specified.

Simplified Approach

Most metropolitan areas will have considerably less detailed data on freight movements and will not have a model structure that allows tracking of freight by commodity class. Even without commodity or truck flow data, it is still possible to estimate user and economic benefits to freight traffic, although additional assumptions and approximations are required. Two possible situations include:

Sources of Commodity Flow Data

The Commodity Flow Survey (CFS), conducted by the Bureau of Transportation Statistics (BTS) and the Bureau of the Census, reports freight tonnage by commodity group between 89 National Transportation Analysis Regions. The CFS data are available free of charge from the Bureau of Transportation Statistics or the Census Bureau.

More detailed commodity flow data can be purchased from Reebie's TRANSEARCH database. This database contains commodity flow data between counties (and in some cases, zip codes), and is based on a more extensive sample and additional information sources compared to the CFS.

An MPO has developed a truck model that produces zone-to-zone truck flows, but does not have data on flows by commodity type. In this case, user benefits to truck traffic can still be estimated in the aggregate using STEAM. To develop inputs of user benefits by commodity type to REMI, commodity flow data are analyzed to identify the mix of commodities by tonnage for the region. Then, data from the Census Bureau's Vehicle Inventory and Use Survey (VIUS; formerly the Truck Inventory and Use Survey, or TIUS) are used to identify average freight volumes (in tonnage) per truck. These two sources are combined to estimate the percentage of truck movements by commodity type, and user benefits from STEAM are then allocated proportionally. This approach assumes that the average benefit per truck is the same across all commodity groups. This would not be the case if different commodity groups follow different travel routes.
An MPO has not developed a truck model. In this case, it is not possible to estimate freight user benefits based on network flows with STEAM. Truck benefits would need to be estimated using a project-specific analysis. If truck travel time benefits can be estimated, a sketch-level methodology can then be used to translate these benefits into REMI inputs for the economic analysis. The Freight Transportation Investment Model (FTIM) provides an example of such an approach. FTIM was developed for Columbus, Ohio but could be applied to any metropolitan area.

[TOP]