Full text not available from this repository.

Abstract

Oil pollution, which is mainly caused by oil tanker accidents, is an important concern in the protection of the marine environment. In order to propose pollution prevention policy for oil tankers, this study first employs four machine learning methods to evaluate oil tanker accident probability, and then proposes a risk evaluation system which is jointly measured by accident probabilities and consequences. Crude oil tanker accident data from 1991 to 2016 were collected and analyzed. The main findings are as follows: First, the random forest (RF) method produces the best overall performance for oil tanker accident probability. Second, the accident probabilities of larger crude oil tankers are lower, while older vessels within the parameter of 0–15 years exhibit declining accident probabilities. Third, tankers classified by the International Association of Classification Societies (IACS) members, registered in closed registry, built in China and South Korea, or owned by developed countries prove to be safer with lower accident probabilities. Fourth, the use of double hull in tankers also contributes to a higher tanker safety level. The results contribute to marine environment protection by helping stakeholders to identify accident risk factors, assess accident risk and take actions to reduce accident probabilities.

Item Type:	Article
Authors/Creators:	Jin, M and Shi, W and Yuen, KF and Xiao, Y and Li, KX
Keywords:	marine environment protection, oil pollution, risk assessment, tanker accident, machine learning method
Journal or Publication Title:	Marine Policy
Publisher:	Elsevier
ISSN:	0308-597X
DOI / ID Number:	10.1016/j.marpol.2019.103655
Copyright Information:	© 2019 Elsevier Ltd. All rights reserved.
Related URLs:	Google Scholar: Shi, W UTAS Profile: Shi, W
Item Statistics:	View statistics for this item

Actions (login required)

Item Control Page