Decision-making on process risk of Arctic route for LNG carrier via dynamic Bayesian network modeling. (July 2021)

• Record Type: Journal Article
• Authors: Li, Zhuang; Hu, Shenping; Gao, Guoping; Yao, Chenyang; Fu, Shanshan; Xi, Yongtao
• Is Part Of: Journal of loss prevention in the process industries. Volume 71(2021)
• Publication Date: 2021-07
• Subjects: Process safety -- Risk decision-making -- Dynamic bayesian network -- Arctic waters -- Maritime traffic risk -- Route planning; Chemical industries -- Safety measures -- Periodicals; 660.2804
• DOI: 10.1016/j.jlp.2021.104473
• Languages: English
• ISSNs: 0950-4230

Abstract

Abstract: An increasing number of ships have chosen the suitable route to transport in Arctic waters during summer. Seeking a suitable model for risk decision-making in route planning is a necessary research topic at present. Due to its complex natural environment, there is significant uncertainty regarding ship navigation safety in Arctic waters. The process risk-based decision-making method to support route planning is established based on the dynamic Bayesian network (DBN) risk assessment model for LNG carrier collision with ice or obstacles in Arctic waters. The decision-making process for ship navigation is dynamically associated with time. Therefore, a Markov Chain (MC) is built for each dynamic node in Bayesian belief network (BBN) to realize DBN associated risk assessment, which is called process risk and is applied to decision-making. Three possible routes for ships sailing from the Vikitsky Strait to the Long Strait in Arctic waters were selected in conjunction with the objective daily change data of wind speed, temperature, wave height, and ice condition. Simulations for risk decision-making in the ship navigation process are performed. Application examples show that the ship selected either ROUTE2 (Vikitsky Strait – Laptev Sea – Sannikov Strait – Eastern Siberian Sea – Long Strait) or ROUTE3 (Vikitsky Strait – Laptev Sea – Proliv Dmitriya Lapteva – Eastern Siberian Sea – Long Strait) in August as the best navigable route. Highlights: From the perspective of uncertain information, modeling research is carried out for risk decision-making problems. A multi-factors network correlation model is proposed to describe the interaction between multiple observation variables of humans, ships, and the environment in the ship transportation system. In view of the uncertain changes of the observed variables during the ship's navigation, the Markov hypothesis is proposed, combined with the spatial state representation of environmental variables, the transition matrix is obtained, and the dynamic Bayesian network model is used to simulate. In view of three choices of the northeast route for LNG carriers to sail in the Arctic waters, combined with the scenario, the route decision-making was conducted on he three months, and the actual analysis conclusions were obtained.