Off-campus WSU users: To download campus access dissertations, please use the following link to log into our proxy server with your WSU access ID and password, then click the "Off-campus Download" button below.

Non-WSU users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Access Type

WSU Access

Date of Award

January 2011

Degree Type

Dissertation

Degree Name

Ph.D.

Department

Industrial and Manufacturing Engineering

First Advisor

Ratna B. Chinnam

Second Advisor

Ekrem A. Murat

Abstract

In just-in-time (JIT) manufacturing environments, on-time delivery is one of the key performance measures for dispatching and routing of freight vehicles. Both the travel time delay and its variability impact the efficiency of JIT logistics operations, that are becoming more and more common in many industries, and in particular, the automotive industry. In this dissertation, we first propose a framework for dynamic routing of a single vehicle on a stochastic time dependent transportation network using real-time information from Intelligent Transportation Systems (ITS). Then, we consider milk-run deliveries with several pickup and delivery destinations subject to time windows under same network settings. Finally, we extend our dynamic routing models to account for arc traffic condition dependencies on the network.

Recurrent and non-recurrent congestion are the two primary reasons for travel time delay and variability, and their impact on urban transportation networks is growing in recent decades. Hence, our routing methods explicitly account for both recurrent and non-recurrent congestion in the network. In our modeling framework, we develop alternative delay models for both congestion types based on historical data (e.g., velocity, volume, and parameters for incident events) and then integrate these models with the forward-looking routing models. The dynamic nature of our routing decisions exploits the real-time information available from various ITS sources, such as loop sensors.

The forward-looking traffic dynamic models for individual arcs are based on congestion states and state transitions driven by time-dependent Markov chains. We propose effective methods for estimation of the parameters of these Markov chains. Based on vehicle location, time of day, and current and projected network congestion states, we generate dynamic routing policies using stochastic dynamic programming formulations.

All algorithms are tested in simulated networks of Southeast-Michigan and Los Angeles, CA freeways and highways using historical traffic data from the Michigan ITS Center, Traffic.com, and Caltrans PEMS.