초록

Abstract ▼

1. Research Purpose
(1) Research motivation
The international trades and demands for global shipping have been increasing due to globalization.
According to this recent trend on the international industry, the burdened amount of shipping freight expects to exceed what shipping industries c

1. Research Purpose
(1) Research motivation
The international trades and demands for global shipping have been increasing due to globalization.
According to this recent trend on the international industry, the burdened amount of shipping freight expects to exceed what shipping industries can handle. However, most of the ports cannot just expand its spatial and numerical capacity due to constraints on land or capital. [1] As a result, the ports will be highly pressurized into more efficient scheduling of dispatch and decision on docking, and shipping industries are endeavoring to investigate optimal shipping scheduling that can improve efficiency and productivity in order to stay competitive in the rapidly changing global trend.

Furthermore, the existence of numerous unforeseen and uncertain circumstances makes the scheduling problem a more complicated dynamic problem. (Figure 1) For instance, the weather condition such as wind and waves closely affect the operational availability and taking time for the vessels to be docked at the port and to load/unload the cargoes. Besides, there are other uncertainty factors significantly influencing the shipping scheduling such as delays in arrival time of vessels to the port, inventory status of products, and fluctuations in the production plan determined in upstream processes. In such a system involving various uncertainties, a deterministic model with some assumptions cannot develop a scheduling strategy that ensures the optimality or even the feasibility.
Therefore, research on stochastic scheduling that provides ‘optimal scheduling decisions under various sources of uncertainty’ is highly important and required. However, detailed studies on the development and application of such a methodology and algorithm for stochastic scheduling are lacking so far.

(출처 : 본문 1. Research Purpose 2p)

목차 Contents

• COVER ... 1
• 1. Research Purpose ... 2
• (1) Research motivation ... 2
• (2) Target system description … shipping system of Borouge, UAE ... 2
• (3) Research competency of PI ... 4
• 2. Research Target ... 5
• (1) Research target of the year ... 5
• (2) Research target of the following years ... 7
• 3. Research Method ... 9
• (1) Hierarchical combination of mathematical programming (MP) and Markov decision process (MDP) ... 9
• (2) Reinforcement Learning: solution algorithm for MDP ... 11
• 4. Research Results ... 13
• (1) Research achievements ... 23
• (2) Further research required ... 23
• References ... 23
• Appendix A. MILP formulation ... 24
• Appendix B. PSCH algorithm ... 28
• Appendix C. Setups of 18 case studies ... 30
• Appendix D. Tramp liner rental planning MDP formulation ... 31
• End of Page ... 33