Maria Hänninen

Influences of variables on ship collision probability in a Bayesian belief network model

The influences of the variables in a Bayesian belief network model for estimating the role of human factors on ship collision probability in the Gulf of Finland are studied for discovering the variables with the largest influences and for examining the validity of the network. The change in the so-called causation probability is examined while observing each state of the network variables and by utilizing sensitivity and mutual information analyses. Changing course in an encounter situation is the most influential variable in the model, followed by variables such as the Officer of the Watchs action, situation assessment, danger detection, personal condition and incapacitation. The least influential variables are the other distractions on bridge, the bridge view, maintenance routines and the officers fatigue. In general, the methods are found to agree on the order of the model variables although some disagreements arise due to slightly dissimilar approaches to the concept of variable influence. The relative values and the ranking of variables based on the values are discovered to be more valuable than the actual numerical values themselves. Although the most influential variables seem to be plausible, there are some discrepancies between the indicated influences in the model and literature. Thus, improvements are suggested to the network.

Bayesian network model of maritime safety management

This paper presents a model of maritime safety management and its subareas. Furthermore, the paper links the safety management to the maritime traffic safety indicated by accident involvement, incidents reported by Vessel Traffic Service and the results from Port State Control inspections. Bayesian belief networks are applied as the modeling technique and the model parameters are based on expert elicitation and learning from historical data. The results from this new application domain of a Bayesian network based expert system suggest that, although several its subareas are functioning properly, the current status of the safety management on vessels navigating in the Finnish waters has room for improvement; the probability of zero poor safety management subareas is only 0.13. Furthermore, according to the model a good IT system for the safety management is the strongest safety-management related signal of an adequate overall safety management level. If no deficiencies have been discovered during a Port State Control inspection, the adequacy of the safety management is almost twice as probable as without knowledge on the inspection history. The resulted model could be applied to performing several safety management related queries and it thus provides support for maritime safety related decision making.

Bayesian networks for maritime traffic accident prevention: Benefits and challenges

Bayesian networks are quantitative modeling tools whose applications to the maritime traffic safety context are becoming more popular. This paper discusses the utilization of Bayesian networks in maritime safety modeling. Based on literature and the authors own experiences, the paper studies what Bayesian networks can offer to maritime accident prevention and safety modeling and discusses a few challenges in their application to this context. It is argued that the capability of representing rather complex, not necessarily causal but uncertain relationships makes Bayesian networks an attractive modeling tool for the maritime safety and accidents. Furthermore, as the maritime accident and safety data is still rather scarce and has some quality problems, the possibility to combine data with expert knowledge and the easy way of updating the model after acquiring more evidence further enhance their feasibility. However, eliciting the probabilities from the maritime experts might be challenging and the model validation can be tricky. It is concluded that with the utilization of several data sources, Bayesian updating, dynamic modeling, and hidden nodes for latent variables, Bayesian networks are rather well-suited tools for the maritime safety management and decision-making.

Bayesian network modeling of Port State Control inspection findings and ship accident involvement

The paper utilizes Port State Control inspection data for discovering interactions between the numbers of various types of deficiencies found on ships and between the deficiencies and ships involvement in maritime traffic accidents and incidents. Bayesian network models for describing the dependencies of the inspection results, ship age, type, flag, accident involvement, and incidents reported by the Vessel Traffic Service are learned from the Finnish Port State Control data from 2009-2011, 2004-2010 Baltic Sea accident statistics and the reported Gulf of Finland Vessel Traffic Service incidents within 2004-2008. Two alternative Bayesian network algorithms are applied to the model construction. Further, additional models including a hidden variable which represents the complete system and its safety features and which links the accident and incident involvement and Port State Control findings are presented. Based on model-data fit comparisons and 10-fold cross-validation, a constraint-based learning algorithm NPC mainly outperforms the score-based algorithm repeated hill-climbing with random restarts. For the highest scoring models, mutual information and influence of evidence analyses are conducted in order to analyze which network variables and variable states are the most influential ones for determining the accident involvement. The analyses suggest that knowledge on ship type, the Port State Control inspection type and the number of structural conditions related deficiencies are among the ones providing the most information regarding accident involvement and the true state of the hidden system safety variable.

Expert elicitation of a navigation service implementation effects on ship groundings and collisions in the Gulf of Finland

When considering the implementation of a novel risk-control option, the estimation of its possible effects often relies on expert elicitation. This article presents an expert-knowledge–based preliminary assessment of how the deployment of Enhanced Navigation Support Information navigation service would affect the ship collisions and groundings in the Gulf of Finland. Experts probabilistically assess the service’s direct effects on various factors, which are then utilized in collision and grounding probability Bayesian network models. The results indicate that implementing the Enhanced Navigation Support Information service could decrease the number of accidents. However, a comparison of the model outcomes to the experts’ qualitative opinions reveals some discrepancies, which suggest that the elicitation procedure or the applied models might require further improvement. Nevertheless, with the proposed Bayesian approach, the model can be updated and uncertainties in the estimates reduced after more evidences are available later from longer and wider use of the service.

Uncertainty in maritime risk analysis: Extended case study on chemical tanker collisions

Uncertainty is inherent to risk analysis. Therefore, it is extremely important to properly address the issue of uncertainty. In the field of risk analysis for maritime transportation systems, the effect of uncertainty is rarely discussed or quantified. For this reason, this article discusses a case study dealing with risk analysis for a chemical spill in the Gulf of Finland and analyses the related uncertainties by adopting a systematic framework. Risk is assessed in terms of the expected spill frequency and spill volumes caused by collisions between ships and chemical tankers in the Gulf of Finland. This is done by applying a collision consequence with a novel approach-to-collision-speed linkage model and Gulf of Finland–specific causation factors, which are based on reanalysing accident data. This article also presents a metamodel for assessing collision probability with initial vessel speeds for any given scenario where a chemical tanker is about to be struck by another vessel. Even when conducting a risk analysis using state-of-the-art methods, there is still a medium-high degree of uncertainty in the model presented in this article, which only becomes apparent when conducting a systematic uncertainty assessment analysis. However, an uncertainty assessment is an important part of quantitative maritime risk analysis. For this purpose, a qualitative framework for uncertainty assessment analysis is introduced for general use in the field of maritime risk analysis.

A Cross-disciplinary Approach to Minimising the Risks of Maritime Transport in the Gulf of Finland

The maritime traffic in the Gulf of Finland is predicted to rapidly increase in the near future, which increases the environmental risks both through direct environmental effects and by increasing the accident risk. This paper describes a multidisciplinary modelling approach, where, based on growth predictions, the maritime traffic in the Gulf of Finland in the year 2015 is modelled and the accident risk, the direct environmental effects and the risk of environmental accidents are evaluated. Finally, the effects of national and international legislation and other management actions are modelled, to produce advice and support for governmental decision makers. In the modelling work, Bayesian Networks (BNs) are applied. The approach produces unique information on the accident risks and their effects separately for each marine route used, which enables efficient local risk control actions to be taken by the decision makers to decrease the probability of accidents.

A probabilistic approach for a cost-benefit analysis of oil spill management under uncertainty: A Bayesian network model for the Gulf of Finland

Large-scale oil accidents can inflict substantial costs to the society, as they typically result in expensive oil combating and waste treatment operations and have negative impacts on recreational and environmental values. Cost-benefit analysis (CBA) offers a way to assess the economic efficiency of management measures capable of mitigating the adverse effects. However, the irregular occurrence of spills combined with uncertainties related to the possible effects makes the analysis a challenging task. We develop a probabilistic modeling approach for a CBA of oil spill management and apply it in the Gulf of Finland, the Baltic Sea. The model has a causal structure, and it covers a large number of factors relevant to the realistic description of oil spills, as well as the costs of oil combating operations at open sea, shoreline clean-up, and waste treatment activities. Further, to describe the effects on environmental benefits, we use data from a contingent valuation survey. The results encourage seeking for cost-effective preventive measures, and emphasize the importance of the inclusion of the costs related to waste treatment and environmental values in the analysis. Although the model is developed for a specific area, the methodology is applicable also to other areas facing the risk of oil spills as well as to other fields that need to cope with the challenging combination of low probabilities, high losses and major uncertainties.

A Bayesian network for assessing the collision induced risk of an oil accident in the Gulf of Finland

The growth of maritime oil transportation in the Gulf of Finland (GoF), North-Eastern Baltic Sea, increases environmental risks by increasing the probability of oil accidents. By integrating the work of a multidisciplinary research team and information from several sources, we have developed a probabilistic risk assessment application that considers the likely future development of maritime traffic and oil transportation in the area and the resulting risk of environmental pollution. This metamodel is used to compare the effects of two preventative management actions on the tanker collision probabilities and the consequent risk. The resulting risk is evaluated from four different perspectives. Bayesian networks enable large amounts of information about causalities to be integrated and utilized in probabilistic inference. Compared with the baseline period of 2007-2008, the worst-case scenario is that the risk level increases 4-fold by the year 2015. The management measures are evaluated and found to decrease the risk by 4-13%, but the utility gained by their joint implementation would be less than the sum of their independent effects. In addition to the results concerning the varying risk levels, the application provides interesting information about the relationships between the different elements of the system.

Marine traffic, accidents, and underreporting in the Baltic Sea

This paper presents an overview of ship traffic volume and accidents in the Baltic Sea with a special focus on the Gulf of Finland. The most common accidents are groundings and collisions, usually reported to be caused by human error. The annual number of Baltic Sea accidents reported to HELCOM varied from 34–54 for collisions and 30–60 for groundings. The number of yearly port calls varied from 468–505 thousand with a peak in 2008. Exact port call data could not be found for all ports and hence had to be estimated. The number of line crossingings in HELCOM AIS data was found to be a good, rough surrogate measure for the total number of port calls and could be used if more precise port call data was not available. By analyzing two separate accident databases, an estimate for accident underreporting was calculated. Different statistical methods yielded an underreporting rate in the range of 40–50%. Lastly, the true number of accidents was estimated, based on the estimated underreporting percentage for the Baltic Sea. Based on these results, the true number of true accidents should be first estimated if accident statistics are used in building or validating maritime risk models. When using such models or accidents statistics in decision-making, the underlying uncertainty in the accident statistics should be taken into account as the underreporting frequency estimates are only approximations of the real number of accidents.