A framework for evidence-based risk modeling of ship grounding

Abstract

Most of the risk models for ship-grounding accidents do not fully utilize available evidence, since in general they are merely based on accident statistics and expert opinions. The major issue with models of such kind is their limitation in supporting the process of risk management with respect to grounding accidents; because they do not necessarily reflect the reality to the extent required.

This thesis proposes an evidence-based framework for building evidence-based risk models for probabilistic assessment of ship grounding. In order to build the evidence required for creating the evidence-based risk model, traffic characteristics as primary source of data are extracted from AIS data and accident statistics of the Gulf of Finland. Additionally, using expert knowledge of the local pilots in the Gulf of Finland, a location dependent and semi-quantitative index as Waterway Complexity Index is defined to assess the dependency of ship grounding and navigational difficulty of a waterway to handle a ship. Moreover, ship grounding incident and accident reports from Finnish, Swedish, and British maritime authorities are utilized as primary and secondary sources of data respectively to build the required evidence for constructing the risk model. In this regard, two frameworks are introduced in this thesis to review and extract the embedded information from the reports. A new version of Human Factors Analysis and Classification System (HFACS) is introduced as a framework to review the grounding accident reports; and a new positive taxonomy as Safety Factors, which are based on high level positive functions that are prerequisite for safe transport operations, is introduced to review the grounding incident reports.

Utilizing the proposed framework for evidence-based risk modeling as well as the built evidence from primary and secondary sources of data and expert knowledge, a Bayesian Network risk model is developed in this thesis for assessing the probability of ship grounding accidents. The uncertainties associated with the elements of the model are clearly communicated to the end user adopting a concept of strength of knowledge, and by introducing knowledge strength map for the built model. Therefore, it is argued in this thesis that the developed model is more suitable for risk management purposes, and the model can be used to suggest proper risk-control-measures to mitigate the risk of ship grounding accident. The developed model in this thesis suggests the high-level critical parameters that need proper control measures are complexity of waterways, traffic situations encountered, and off-coursed ships. The critical area that calls for more investigation is the onboard presence of a sea-pilot.