A trip-based travel demand framework consistent with tours and stop interaction.

摘要

Beginning in the 1960s, U.S. federal law required ongoing and systematic planning as a condition of federal transportation funding. An ad-hoc system for forecasting future travel, which has come to be referred to as the "four-step" procedure, quickly became established as the ubiquitous analytic tool for transportation planning. The four-step procedure has long been criticized for a variety of shortcomings and inconsistencies. In the past ten years, an alternative activities-based approach has been adopted by a handful of planning agencies across the country. However, the cost and complexity of the activities-based approach, the only substantial alternative, has thus far prevented its widespread adoption.;This dissertation presents a new system for forecasting future travel which addresses many of the critiques of the traditional four-step procedure with significantly less complexity and greater efficiency than the activity-based framework. The core of this method is a two-stage model of the choice of stop locations and stop sequences or trips between them. This approach addresses the most glaring inconsistency of the traditional procedure by ensuring that all travel is consistent with closed tours, or continuous space-time trajectories which return to the residence location where they began. The ability of this model to ensure consistency with tours without enumerating the tours themselves is the key source of its simplicity and cost savings compared to activity-based techniques. This approach offers an attractive option particularly for regions with very little transit use and limited resources for developing advanced planning models.;Beyond the basic inconsistency with tours, a lack of consistent sensitivity to both radial and angular dimensions of the distribution of stop locations around the home is proposed as a scheme for critiquing the spatial distribution of travel demand predicted by traditional models. A new class of destination choice models, extending Fotheringham's competing destinations model, is developed to incorporate, in addition to differential spatial competition effects, agglomeration effects related to the angular dimension of the distribution of stops. Empirical research with these models offers the first exploration of the scope as well as magnitude of these two stop interaction effects.