| Most academic studies in the field of traffic engineering focus on either multiclass-user or multi-mode traffic assignment problems and are interested in the convergence of the computation and the uniqueness of the equilibrium. On the other hand, most practical studies focus on the travel predictions and mode-selection behavior. Therefore, a model that describes the mode-selection for multiple classes of users is needed to bridge the multiclass-user approach with multi-mode considerations. In this extended structure, the computation task is complex and assumptions about the travelers' behavior have more constraints.; The research presented in this dissertation models the real-world travelers' decision behavior considering multiple income classes, multiple transport modes, and physical network constraints. By introducing an intangible cost (the mode-aversion cost) into the generalized travel cost function and by applying fixed-point theory from nonlinear complementary programming, this research successfully extends Wardrop's user optimal principle to transport mode selection, which results in a more realistic simulation.; Travelers' decisions result in traffic congestion, air pollution, and other social costs for the whole urban area. To develop a sound transportation policy, the transportation planners need to fully understand the travelers' behavior and their reactions to the policy. Therefore, a model needs to be developed that matches these real-world situations to help the transportation planners simulate various traffic situations and to predict the economic and social impacts of transportation policies.; One of the main purposes for this dissertation research is to provide a system-integrated analytic model for transportation planners and decision-makers. This model shows the impacts of transportation policies from a system point of view, which considers the total travel time, total travel costs, fuel consumption, and CO emission. This model was applied to the case of Taipei city to test transportation policies that are being (or should be) considered. The results of this research demonstrate the model's ability to predict the transportation situations and transportation-related social impacts. This model readily allows researchers to test most of the transportation policies (for example, the fuel price policy, parking/toll policy, and transit fare policy) by configuring a few parameters. |
