Kıyı terminali operasyonları için matematiksel modeller

Maritime terminals are the key components of global freight transportation as they handle over 80\% of global trade by volume and more than 70% of value according to United Nations. A steadily increasing workload causes maritime terminals around the world to face with a high competitive pressure. Therefore, it is essential for them to improve their performance levels in terms of service rate and costs. Maritime terminals can be classified into two based on the transported materials: container and bulk. Differently from container terminals in which standardized containers are processed, bulk terminals deal with unpackaged natural resources and agricultural products, such as iron ore, coal, grains, oil, and gas in large quantities. Operational problems observed in both classes of maritime terminals are the variants of well studied operations research problems such as machine scheduling, vehicle routing, and bin packing. However, they are more complex in nature because of the distinctive characteristics of terminals. In this dissertation, we study three different but related planning and scheduling problems of maritime terminals with a practical relevance: quay crane assignment problem (QCAP), reclaimer scheduling problem (RSP) and integrated dry bulk terminal (IDBT) problem. In each of these problems, we deal with one of the most challenging characteristics which is caused by the fact that bottleneck equipment in both classes of terminals, namely, quay cranes (QC) and reclaimers, are mounted on the same rail track, thus their movements are restricted by their respective positions over time. In the first chapter, we introduce container and bulk terminals by presenting a concise overview of their operations as well as the related literature. Within these sections, we also describe our motivations and contributions for QCAP, RSP, and IDBT problem, respectively. In Chapter 2, we study QCAP. In this problem, QCs are assigned to arriving vessels and handling time of a vessel depends on the number of assigned QCs. As QCs are the bottleneck equipment in container terminals, they need to be utilized efficiently. We study a QCAP by considering its various features, differently from the literature which deals with the simplest form of the problem. With our approach, we can estimate the vessel handling times accurately and hence improve the productivity as a result. We represent this problem as a moldable task scheduling problem with contiguous assignments, and we formulate an extended time-indexed model that additionally keeps specific task to machine information. Even though extended formulation is flexible in terms of modeling different features of the problem, it has multiple computational issues. Therefore, we develop an exact solution method by first hybridizing the formulation with a set of auxiliary variables to obtain a decomposable structure, and then implementing logic-based Benders decomposition (LBBD) with strong cuts. One limitation of this proposed method is excessive memory requirements for large instances, since it is based on a time-indexed formulation. Hence, we propose methods to derive lower and upper bounds for larger problem instances in Chapter 3. Since linear programming relaxations of time indexed formulations are known to be very tight, we developed a column generation procedure in which pricing problem can be solved in polynomial time. For an upper bound, we introduce a constraint programming model that finds near optimal solutions in a short time. In Chapter 4, we introduce the reclaimer scheduling problem. Reclaimers handle the dry bulk cargo, which is stacked as a stockpile. Reclaimers are mounted on the multiple rail tracks. If there are two reclaimers on the same rail track, then they cannot cross each other. Furthermore, a reclaimer can only handle stockpiles located adjacent to its rail track. For this strongly NP-hard problem, we opt for a heuristic approach by developing an arc-time-indexed lower bound model and constraint programming model to find near optimal solutions. There are many relations between operational problems in maritime terminals. If we solve these problems hierarchically, we often end up with plans with poor overall quality. Accordingly, in Chapter 5, we study the integrated problem dry bulk terminal operations, by considering berth allocation, yard assignment, and reclaimer scheduling operations simultaneously. After observing a key relation between problems, we decompose the problem into two easier problems and solve with a novel LBBD. In this decomposition, we model master and subproblems with mixed-integer programming and constraint programming, respectively, by exploiting the respective advantages of programming paradigms. Results show that the proposed method is able to solve considerably large instances to optimality in an acceptable time, compared to the monolithic approach. We conclude the dissertation with Chapter 6 in which we present a concise overview of our contributions and discuss future research directions.