| Urban transportation systems are basic components of cities' social, economic and physical structure. The design and planning of these systems have been traditionally aimed at improving mobility, modifying growth patterns and enhancing economic development; though in more recent years they have been redirected to also serve other national and community objectives (e.g., social equity, sustainability). To make this possible, urban planners count on a number of tools to anticipate the consequences of alternative actions. In particular, transportation demand models play an important role in support planning by providing a simplified representation of a real world problem. In the case of freight, the complexity of interacting decision makers, the multiple dimensions to measure freight flows, the lack of appropriate knowledge, and the availability of suitable data, represent a threat for applying the classical modeling approaches (i.e., 4 steps model). These issues within freight transportation systems have unveiled the need for new freight urban demand models.;This dissertation proposes a methodology for estimating a demand model that considers the tour-based behavior of trucks, the temporal aspect, and simultaneously incorporates data from secondary data sources (i.e., traffic counts) to obtain freight vehicles flow distribution. The inputs to the model require a tour generation model, the application of freight trip generations models, and of a prior traffic assignment model. The flow distribution analytical model is a bi-objective optimization problem, in which the first objective minimizes the square differences between observed and estimated link flows; while the second objective is part of on an entropy maximization (EM) formulation that finds the most likely flow distribution satisfying a set of constraints (i.e., demand and total impedance of the network). The resulting Pareto frontier reveals the tradeoffs between EM and minimizing traffic estimation's error. The two objective functions are then merged using micro-economics principles, resulting in a multi-attribute value function. In essence, the Time-Dependent Freight Tour Synthesis (TDFTS) proposes an innovative methodology to incorporate the data from traffic counts into an entropy-based problem, while respecting the preference structure of the decision maker (e.g., a Metropolitan Planning Agency planner).;The Metropolitan Region of Denver (Colorado) is proposed as a case study to test the model, produce an assessment of the results, and perform a sensitivity analysis. The results demonstrate the great potential of the TDFTS to reproduce freight vehicle flows even when they are differentiated by industry and time interval. |
