Date of Award:


Document Type:


Degree Name:

Doctor of Philosophy (PhD)


Civil and Environmental Engineering

Committee Chair(s)

Anthony Chen


Anthony Chen


Ziqi Song


Gilberto E. Urroz


Haitao Wang


Yong Seog Kim


The current practice for modeling in the field of transportation planning is through a four-step travel demand forecasting procedure (i.e., trip generation, trip distribution, mode choice, and traffic assignment); the practice is commonly referred to as the four-step model. Although such a modeling approach has become standard practice, it is deficient in several areas. Specifically, (1) it lacks capability for modeling non-motorized modes such as bicycles, (2) it is inadequate for modeling multiple vehicle types sharing the same roadway space, and (3) it is difficult to apply to small communities with limited resources. This dissertation recognizes these deficiencies and responds to them through the development of alternative transportation planning applications via the path flow estimator (PFE). The PFE was originally developed as a one-stage network observer capable of estimating path flows and path travel times using only traffic counts from a subset of network links. In this dissertation, the PFE is used to develop the following three transportation planning applications for addressing the three deficiencies of the four-step model: (1) a bicycle network analysis tool for non-motorized transportation planning, (2) a multi-class traffic assignment model for freight planning, and (3) a simplified travel demand forecasting model for small community planning.

The first application develops a two-stage bicycle traffic assignment model for estimating/ predicting bicycle volumes on a transportation network. The first stage
considers two key criteria (e.g., distance related attributes and safety related attributes) to generate a set of non-dominated (or efficient) paths, while the second stage determines the flow allocation to the set of efficient paths. In stage one, a bi-objective shortest path problem based on the two key attributes is developed to generate the efficient paths. In stage two, several traffic assignment methods are adopted to determine the flow allocations in a network. In addition, the two-stage bicycle traffic assignment model is further extended to consider multiple user classes and multiple criteria. This two-stage approach can be used as a stand-alone bicycle traffic assignment to the transportation network given a bicycle origin-destination (O-D) matrix.

The second application aims to enhance the realism of traffic assignment models for freight planning by incorporating different modeling considerations into the multi-class traffic assignment problem. These modeling considerations involve developing both model formulation and customized solution algorithm that in turn, involve asymmetric interactions among different vehicle types (i.e., cars versus trucks), a path-size logit (PSL) model (for accounting random perceptions of network conditions with explicit consideration of route overlapping), and various traffic restrictions imposed either individually or together to multiple vehicle types in a transportation network. Specifically, a variational inequality (VI) approach is used to formulate the stochastic multi-class traffic assignment problem with asymmetric vehicle interactions, route overlapping, and traffic restraints. In addition, a customized path-based solution algorithm, consisting of an iterative balancing scheme, a self-regulated averaging line search scheme, and a column generation scheme, is developed for solving the stochastic multi-class traffic assignment problem.

The third application develops an alternative planning tool to model and forecast network traffic for planning applications in small communities, where resources debilitate the development and applications of the conventional four-step travel demand forecasting
model. Two versions of the PFE are developed to address the specific transportation planning issues and needs of small communities: a base year PFE for estimating the current network traffic conditions using field data and planning data if available and a future PFE for predicting future network traffic conditions using forecast planning data and the calibrated origin-destination trip table as constraints. Solution algorithms are also developed to solve the two PFE models; they are integrated into a GIS-based software tool called Visual PFE, which is streamlined for planning applications in small communities.

To show proof of concept, these new PFE developments for planning applications are tested with different realistic transportation networks. The results suggest that the new
PFE applications developed in this dissertation provide an alternative to the traditional four-step travel demand forecasting model that can be used as a stand-alone application with better modeling capability and fewer resources.