یک مدل برنامه ریزی پویا، مبتنی بر سناریو برای برنامه ریزی ناوگان کشتی لاینر چند دوره ای

This paper proposes a more realistic multi-period liner ship fleet planning problem for a liner container shipping company than has been studied in previous literature. The proposed problem is formulated as a scenario-based dynamic programming model consisting of a number of integer linear programming formulations for each single planning period, and the model can be solved efficiently by a shortest path algorithm on an acyclic network. A numerical example is carried out to illustrate the applicability of the proposed model and solution method. The numerical results show that chartering in ships may not always be a better policy for a long-term planning horizon though it is much cheaper than buying ships in the short-term. Purchasing ships seems to be a more profitable investment in the long run.

The increasing globalization and inter-dependence of various world economies are leading to a significant positive growth in seaborne trade. According to the review of maritime transport reported by the United Nations Conference on Trade and Development (UNCTAD) secretariat, international seaborne trade increased from 2566 million tons in 1970 to 8168 million tons in 2008 (Chapter 1 of UNCTAD (2009)). In particular, containerized trade by liner shipping is the fastest growing sector in global seaborne transportation, as a result of a combination of various factors, including dedicated purpose-built container vessels, larger vessels capable of achieving increased economies of scale, improved handling facilities in ports, and also the increasing amount of raw materials being carried in containers (here a container refers to the 20-foot equivalent unit [TEU]). With the continuous advancement of ship-building technology and the increase in global container traffic, the dominance of liner shipping is expected to continue to strengthen (Chapter 4 of UNCTAD (2009)). In order to seize market share in an intensely competitive container shipping market, a liner container shipping company has to make a suitable strategic fleet development and management plan. Therefore, it is critical for such a company to project its fleet. The multi-period liner ship fleet planning (MPLSFP) problem aims to determine the optimal joint fleet development and deployment plan, within a multi-period planning horizon, for a liner container shipping company to effectively provide shipping services. A fleet development plan is used to identify the current and future types and quantities of ships required, while a fleet deployment plan is used to determine the assignment and operation of the fleet to transport containers. A joint fleet development and deployment plan thus includes chartering, purchasing, selling, assigning and operating ships. Hence, this paper focuses on model development and algorithm design for the MPLSFP problem with the objective of profit maximization subject to a given container shipment demand.

This paper has investigated a more realistic MPLSFP problem than has been studied in previous literature. In order to capture the characteristics of a realistic dynamic decision strategy, the multi-period planning horizon was divided into a number of single periods and a scenario decision tree was introduced to formulate the determination procedure in the MPLSFP problem. The MPLSFP problem was thus formulated as a scenario-based dynamic programming model. The proposed scenario-based dynamic programming model is equivalent to a longest path problem in an acyclic network. Hence, it can be solved by any shortest path algorithm. In effect, the proposed scenario tree simulates the decision making process of a multi-period liner ship fleet plan that would be made in practice by a liner container shipping company. The most significant contribution of this paper to the literature is that the methodology developed in this paper provides an applicable and feasible way for a liner container shipping company to make a multi-period liner ship fleet plan that would work in practice. It should be noted that the container shipment demand is assumed to be deterministic and known in our model. It may be uncertain in practice and therefore we intend to make the MPLSFP problem with uncertain container shipment demand the subject of a future study.