Browse Theses

Ships transport 90% of the world's goods, and without shipping, "half the world would starve and the other half would freeze!" according to a well-known adage. To accommodate the steadily increasing demand for shipping services, it is essential to efficiently utilize seaport resources and to optimize the respective operations. This dissertation addresses this important problem. Specifically, it addresses optimization modeling for a number of port seaside operations. Problems considered involve the design of mathematical formulations, the design of solution methodologies that can find optimal or close-to-optimal solutions in a reasonable time, the conduction of computational analyses to assess the performance of the proposed techniques, and the conduction of real and simulated case studies to provide managerial insights.The Quay Crane Scheduling Problem (QCSP) with non-crossing and safety clearance constraints is first considered. The problem determines the order of discharging and loading operations that a specific number of quay cranes perform to serve a vessel in minimum time. Due to the difficulty of this problem, most researchers have used heuristics to solve it. While the QCSP is normally used as a building block in a larger optimization problem addressing port operations, these optimization problems are difficult to solve without decomposition techniques like Lagrangian relaxation. For these methods to succeed, the sub-problems must be solved to optimality in reasonable computational time. We introduce an improvement on a recent novel formulation for the problem, followed by a new exact and computationally fast technique to solve it. The technique is a two-step approach initiated by a partitioning heuristic and terminated by a branch-and-price algorithm. Through computational experiments, we demonstrate that the proposed solution approach can solve real-sized cases efficiently and with low sensitivity to all parameters.Next, berth allocation and quay crane assignment decisions are integrated into the QCSP. We introduce a new mathematical formulation that captures all associated operations and constraints. Different quay crane operational policies are considered, namely permitting versus not permitting bay task preemption and static versus dynamic crane allocation. Variants of the mathematical formulation are introduced to capture the different combinations of these scenarios. Due to the preemption consideration, the models include disaggregated quay crane tasks. Specifically, quay crane tasks are identified by single-container movements as opposed to the bay or stack task allocations that are commonly seen in the literature. A case study based on Abu Dhabi's container terminal is presented in which the proposed mathematical models are compared against the operational approach that is currently in use. The results show that service times can be significantly decreased by the use of the proposed models. Moreover, the effect of policy choice on the total schedule is compared through simulated examples and a case study based on Abu Dhabi's container terminal.We next consider the vessel scheduling problem in the berthing context. Vessel scheduling has a direct impact on vessel waiting times and berthing operation completion times. Thus, their optimization is essential to achieving higher customer satisfaction and better resource utilization. To our knowledge, berthing resources have not been considered in the context of vessel scheduling with channel restrictions, and existing solution approaches use approximation techniques. We propose a mixed integer program (MIP) and an exact solution approach, based on constraint separation techniques, to solve the problem. The solution approach is initiated with a heuristic that runs in polynomial time. The approach then adds separation cuts to a relaxed version of the MIP and iteratively improves the bounds until an optimal solution is found. A real case and simulated realistic cases are used to compare the proposed model against the first-come first-serve policy that is traditionally used in vessel scheduling. The results show that significant improvements can be realized, especially during congestion periods. Next, a resource sensitivity analysis is presented that sheds light on how the model can be used for resource purchasing decisions. Finally, a computational study demonstrates that the proposed solution approach is capable of solving real-sized cases in a reasonable time.Finally, scheduling of the recently patented next-generation quay cranes is considered. These cranes can access two bays simultaneously and can operate on four containers at a time. We first consider the use of next-generation cranes in isolation, and then consider the next-generation cranes operating in conjunction with traditional ones. We introduce mixed integer programs and solution approaches to solve each of the problems. The solution technique for the case of next-generation cranes working in isolation breaks the main problem into two sequential stages. The first stage uses a fast set-partitioning formulation to solve the general case and a closed-form analytic approach to solve specific cases, while the second stage uses a partitioning heuristic combined with a branch-and-price algorithm. The solution technique for the case of traditional and next-generation cranes working side-by-side uses a column generation algorithm to solve a re-visualized vessel structure. The new structure's workload is enumerated using a fast set-partitioning formulation. Case studies are conducted to assess the performance of the next-generation cranes and to shed light on how the positioning of these cranes affects the service time. The results show that average service times can be reduced by up to 65%. Through computational studies, we show that the proposed approaches can solve real-sized cases efficiently.