Land Use and Travel Demand Forecasting Models

In April 2003, the Regional Council completed a project to update its travel demand model. The final reports are now available:

Download Download earlier reports

In August 2000 the Regional Council hired Cambridge Systematics to document and validate its travel demand models, and to make recommendations for both short term and long term improvements to those models. The work was divided into five tasks:

  • Task 1 - Document the various models
  • Task 2 - Validate the current models
  • Task 3 - Validate the models with two new components previously developed but never made operational
  • Task 4 - Define a complete 2000 database for future model development
  • Task 5 - Recommend short term and long term improvements

In addition, in March 2001 the Washington State Department of Transportation (WSDOT) requested and paid for an additional task in which the consultants would clarify and respond to concerns of legislative critics of the models.

The validation of the current models shows that the various model components are generally within professionally accepted standards for accuracy, though there is room for improvement. Where they exist, deviations from observed data tend to offset each other, so the net result is a reasonable representation of travel throughout the region.

The validation of the models, using the new components for trip distribution and mode choice, shows that the new models are not ready for general use. However, since the current models and the new models share trip generation and assignment components, and since the new models offer more features for evaluating a wider range of policy questions, the consultants recommend basing the short term model improvements on the new models.

The consultants are recommending short term improvements to the new models, and the mid-term and long term development of a new way of modeling travel behavior.

Short term improvements (1-2 years) include:

  • Review and update input data assumptions, including the coding of transit and roads, and updating costs and fares to 2000 dollars.

  • Expand the time-of-day model to five time periods (early a.m., a.m. peak, mid-day, p.m. peak, and evening).

  • Update external trip tables. This will involve conducting origin-destination surveys at select cordon locations to determine the movements of trips passing through external stations.

  • Update trip assignment parameters and assumptions, including refined volume-delay functions, generalized cost to better represent fares and tolls, better representation of the usage of HOV lanes, capacity constraints on park-and-ride lots, more accurate representation of cross-Sound trips, and explicit inclusion of trucks and buses.

  • Incorporate the freight model developed as part of the FAST Phase II study to model the traffic impacts of light-, medium-, and heavy-duty trucks.

The cost of the short term improvements would be $250,000 to $350,000 over 1-2 years, which is included in the PSRC's current budget.

Mid-term changes (3-4 years) to the modeling system include:

  • Replace the current land use and travel demand models with an integrated land use and forecasting model. This new model would replace all components of the current models except auto and transit assignment. It would integrate long-term household choices of residential location, workplace, and vehicle ownership with short-term choices of daily activity and travel. Models of this type have been developed in San Francisco and Portland and are currently being tested.

  • Add a visitor trip model to the resident travel model. This will involve conducting a travel survey of visitor behavior.

  • Develop institutional partnerships to fund, develop, implement, and maintain the new model.
The additional cost of the mid-term improvements would be $1,500,000 to $2,500,000 over the next four years.

Long term changes (5-10 years) include:

  • Replace the traffic assignment component with dynamic microsimulation.

  • Adapt the activity-based travel model to continuous time representation.

  • Expand the model to include weekends. This would involve conducting a household survey of weekend travel behavior.

  • Adapt the regional economic model to a multi-region model of commodity-freight transport.

The additional cost of the long term improvements would be $2,000,000 to $3,000,000.

The Regional Council, through its Regional Technical Forum, has already initiated discussions with its members about partnerships in the development of a new model structure such as that recommended for the mid-term. Funding sources beyond those currently identified in the budget are being explored.

Reports

There are eleven reports which present the results of these tasks. All are available below in PDF format. A CD-ROM containing all reports is available upon request from the Information Center, info@psrc.org, 206-464-7532.

Land Use and Travel Demand Forecasting Models:

For further information about these reports and the model improvement project, contact Larry Blain, lblain@psrc.org, 206-971-3287.