Integrated Dynamic Travel Models: Recent SHRP2 Projects
Corresponding Author: Scott Smith, Volpe Center / US DOT
Presented By: Scott Smith, Volpe Center, US DOT
Abstract
Transportation planners are increasing their use of activity based modeling (ABM) and dynamic traffic assignment (DTA) to model travel behavior, congestion, and responses to network changes, at a high level of spatial and temporal detail.
An important next step is to build an integrated model that links travel behavior with congested network conditions to better reflect real-world dynamics. In late 2014, the Second Strategic Highway Research Program (SHRP2) made four implementation assistance awards to build integrated models, and one award to apply the methods of the SHRP2 C04 project, “Improving Our Understanding of How Highway Congestion and Pricing Affect Travel Demand.” Atlanta and Ohio are integrating the CT-RAMP ABM with the DynusT DTA model. Maryland is integrating both an ABM and an agent-based model with DTALite. San Francisco and Seattle are integrating ABM with dynamic transit passenger assignment using the Fast-Trips model. These projects are now nearing completion. Finally, San Diego used ideas from C04 to improve the sensitivity of its existing travel demand model to congestion, travel time reliability, and pricing.
In this workshop, we will hear from the teams about their motivations for using advanced models, the contribution that these projects have made, and the challenges that they faced. Topics will include:
• Applications of integrated models. Agencies are now considering operational actions, including dynamic tolling, to address congestion issues. Thus, our models need to have a strong behavioral basis, and be sensitive to these operational policies.
• Incorporation of reliability and pricing. It is necessary to establish a distribution of value-of-time, and to address the challenge of characterizing trip-level reliability.
• Networks and traffic controls. Detailed models call for detailed networks, including an accurate representation of traffic controls.
• Model integration. With more detailed spatial and temporal resolution, and multiple values of time, the number of skims can become enormous. An alternative approach uses individual trajectories. We will also talk about schedule adjustments, convergence and the relationship with agent-based modeling.
• Applications to transit. We address the challenge of a large, congested, unreliable transit network that presents many potential paths for a trip, paths that may change from day-to-day.