Application of Activity-based Models for Highway Pricing Studies in the U.S.
P Vovsha, J Freedman, Parsons Brinckerhoff, USA
Highway pricing is a transportation policy where Activity-Based Models (ABMs) have clear and tangible advantages over 4-step models. The paper summarizes recent experience in application of ABMs for various pricing studies in the U.S.
Highway pricing is a transportation policy where Activity-Based Models (ABMs) have clear and tangible advantages over 4-step models. The paper summarizes recent experience in application of ABMs for various pricing studies in major metropolitan areas in the U.S. including San-Francisco, New York, and Chicago. These regions already have toll facilities that allow for statistical analysis and estimation of impacts of congestion and pricing on travel demand. In these regions, a wide range of new large-scale pricing schemes was recently explored that provided an extensive material for model sensitivity testing. The discussion is organized by the following main topics:
- User segmentation in the demand model and highway network procedures. One of the primary advantages of a microsimulation ABM is a practically unlimited population and travel segmentation. It is essential for pricing studies where different segments may have very different willingness to pay for travel time savings and reliability improvements. It is shown that 9-10 travel segments combined with 7-8 population segments provides a reasonable level of segmentation that can also be supported in model estimation. In aggregate network simulations, a parallel level of segmentation can be supported in multi-class assignments but the classes have to be grouped by value of time rather than travel purpose or person type.
- Distributed Value of Time (VOT) and other behavioral parameters. Within each segment, willingness to pay is subject to a significant variation across individuals and situations. A model that operates with discrete VOTs may exhibit illogical abrupt responses to small changes in toll values. Microsimulation framework allows for an effective randomization of VOT as well as other parameters across individual agents within the same segment. For example, the technique of individual parameter variation has been successfully applied for modeling license plate rationing.
- Travel dimensions affected by congestion and pricing. Pricing directly affects tour-level and trip-level choices (route, mode, destination, and time of day) through the generalized cost component in the utility functions. Recently included advanced feature of ABMs allows for capturing impacts of congestion and pricing on activity generation and car ownership choice through a wide range of accessibility measures.
- Time-of-day choice with a fine temporal resolution. Modeling impacts of congestion and pricing (and peak spreading effects in particular) requires a level of temporal resolution of 30 min or less. This results in a large number of alternatives (thousands) when multi-dimensional tour-scheduling choices are modeled. In the recent ABM versions, special combinatorial methods for treatment these choices were applied. The developed approach ensures a full consistency of individual daily schedules without gaps or overlaps in the sequence of activities and trips between them.
- Reliability and perceived highway time by congestion levels. Travel time reliability has been recognized as a very important decision-making factor along with average travel time and cost. Simplified indirect measures as differential time perception by congestion levels (proxy for reliability) have been already included in several ABMs. Several new methods to include direct measures of reliability like Standard Deviation of travel time are discussed.
- Treatment of vehicle occupancy. Vehicle occupancy is a very important factor that strongly affects willingness to pay as well as frequently used for eligibility to use Managed Lanes (HOV/HOT lanes). Two different modeling approaches that have been used in applied ABMs will be compared. The first one is simpler and considers occupancy as part of individual mode choice. The second one is more complex and based on an explicit modeling of joint travel as a separate travel segment.
- Route type choice integrated with mode choice. All-or-nothing route choice framework embedded in a deterministic traffic assignment has inherent drawbacks in portraying the proportion between those who chose a tolled route and those who do not. In this regard adding an explicit choice of route type (toll vs. non-toll) as the lower level in the mode choice structure helps compensate for this deficiency. Recently, a similar problem has been recognized with respect to other types of Managed Lanes (not necessarily tolled). Hence route type choice has been extended to incorporate distinctive route types explicitly.
Association for European Transport