| The trucking industry is vitally important to the economy. The less-than-truckload (LTL) industry is an important segment, serving businesses that ship quantities ranging from 150 lbs to 10,000 lbs.; Chapters II and III describe a scheme for the dynamic scheduling of linehaul drivers developed for one of the largest LTL carriers in the United States. Dynamic driver scheduling is challenging because driver schedules must satisfy a complex set of rules, most of which are specified by government regulation to ensure highway safety. In addition, trucking moves, unlike commercial airline flights or train dispatches, are not pre-scheduled; typically, a truck is dispatched when a sufficient amount of freight has accumulated at a terminal and truck capacity can be utilized effectively. The technology developed in this dissertation combines greedy search with enumeration of time-feasible driver duties, and is capable of generating cost-effective driver schedules covering 15,000--20,000 loads in a matter of minutes. The driver schedules satisfy a variety of real-life driver constraints, including U.S. DOT hours-of-service requirements (Chapter II) and union rules (Chapter III). Computational experiments justify the algorithmic design choices, and a comparison with real-world dispatch data from the LTL carrier indicates that the technology produces high-quality driver schedules.; Chapter IV describes a tactical tool for determining the allocation of drivers in a trucking terminal network. One of the key tactical questions faced by an LTL carrier is how many drivers to locate (or domicile) at each terminal. Determining an effective driver allocation can be especially difficult due to union rules. Most carriers are unionized and a portion of their drivers, called bid drivers, can only move loads between their domicile (home location) and a designated region. The driver allocation technology developed determines the number of drivers to allocate to each terminal as well as the designated region for each bid driver. Computational experiments using truck movement data representative of operations at a major U.S. LTL carrier demonstrate the effectiveness of our domiciling methodology, and show that union restrictions may result in substantially larger driver fleets, in some cases even up to 50% larger.; Chapter V, the final part of the thesis, investigates questions related to the number of drivers required to fulfill a given set of loaded truck moves in a more academic setting in order to obtain some fundamental insights. To facilitate the analysis, some simplifying assumptions are introduced: the terminal network consists of only two terminals and the exact dispatch times of the loaded moves are known. The goal is to determine the minimum number of drivers required to cover all the loaded moves, and the resultant dispatch schedule for these drivers. (Abstract shortened by UMI.)... |
