Date of Award


Document Type

Open Access Dissertation


Civil and Environmental Engineering


College of Engineering and Computing

First Advisor

Nathan Huynh


In the last decades, intermodal freight transport is becoming more attractive in the global supply chains and freight transport policy makings. Intermodal freight transport provides a cost-effective, reliable, and efficient movement of freight by utilizing the strengths of different transport modes. The initial and final segment of intermodal freight transport, performed by truck, is known as “drayage.” The scheduling of truck movements in drayage operation within the service area of an intermodal terminal is an operational problem which leads to a truck scheduling problem that determines the efficient schedule of trucks while satisfying all transportation demands and constraints. Drayage accounts for a large percentage of the origin-destination expenses in the intermodal transport. Efficient planning of the drayage operations to improve the economic performance of this operation can increase the efficiency and attractiveness of intermodal transport. The primary objective of this research is to apply operation research techniques to optimize truck movements in drayage operation.

The first study in this dissertation considers the drayage problem with time constraints at marine container terminals imposed by the truck appointment system and time-windows at customer locations. A mathematical model is proposed that solve the empty container allocation problem, vehicle routing problem, and appointment booking problem in an integrated manner. This model is an extension of a multiple traveling salesman problem with time windows (m-TSPTW) which is known to be NP-hard (i.e., non-deterministic polynomial-time hard). To solve this model, a reactive tabu search (RTS) algorithm is developed and its accuracy and computational efficiency are evaluated against an industry-established solver IBM ILOG CPLEX. In comparison with the CPLEX, RTS was able to find optimal or near-optimal solution in significantly shorter time. This integrated approach also allows for more accurate evaluation of the effects of the truck appointment system on the drayage operation.

The second study extends the drayage literature by incorporating these features in drayage problem: (1) treating tractor, container, and chassis as separate resources which are provided in different locations, (2) ensuring that container and chassis are of the same size and type, (3) considering the possibility that drayage companies can sub-contract the work to owner-operators, and (4) a heterogeneous mix of drayage vehicles (from company fleet and owner-operators) with different start and end locations is considered; drayage company’s trucks start at company’s depot and should return to one of the company’s depots whereas owner-operators’ trucks should return to the same location from where they originated. A mixed-integer quadratic programming model is developed that solves scheduling of tractors, full containers, empty containers, and chassis jointly. A RTS algorithm combined with an insertion heuristic is developed to tackle the problem. The experimental results demonstrated the feasibility of the developed model and solution methodology. The results show that the developed integrated model is capable of finding the optimal solutions and is solvable within a reasonable time for operational problems. This new model allowed us to assess the effectiveness of different chassis supply models on drayage operation time, the percentage of empty movements and air emissions.

The fourth work builds on our previous work and extends the integrated drayage scheduling model to consider uncertainty in the (un)packing operation. Recognizing the inherent difficulty in obtaining an accurate probability distribution, this paper develops two new stochastic drayage scheduling models without explicit assumption about the probability distributions of the (un)packing times. The first model assumes that only the mean and variance of the (un)packing times are available, and the second model assumes that the mean as well as the upper and lower bounds of the (un)packing times are available. To demonstrate the feasibility of the developed models, they are tested on problem instances with real-life characteristics.

Future work would address the real-time scheduling of drayage problem. It would assume trucks’ locations, travel times, and customer requests are updated throughout the day. We would propose a solution approach for solving such a complex model. The solution approach would be based on re-optimization of the drayage problem and consist of two phases: (1) initial optimization at the beginning of the day, and (2) re-optimization during operation.

The third study of this dissertation addresses the impact of a new trend in the North American intermodal terminals in using second-tier facilities on drayage operation. These facilities are located outside the terminals and are used to store loaded containers, empty containers, and chassis. This work builds on our previous work and extends the integrated drayage scheduling model to incorporate these features into drayage problem: (1) trucks do not have to wait at customers’ locations during the packing and unpacking operations, (2) drayage operations include a drop yard (i.e., second-tier facility) for picking up or/and dropping off loaded containers outside the marine container terminal, and (3) the job requests by customers is extended to include empty container pickup, loaded container pickup, empty container delivery, and loaded container delivery. As the mathematical model is an extension of the m-TSPTW, a RTS combined with an insertion heuristic developed by the authors is used to solve the problems.