On Human Reliability Analysis In Collision Avoidance Of Ships

Maritime accident statistics show that the rate of collision of ships is always ranking in front of the all kinds maritime accidents. The consequences of collision accident are also disastrous. Research confirms that more than 90% of the collision accidents are fully or partly caused by human errors.The fact that human elements are involved deeply in maritime accidents has drawn the attention from the international communities. IMO, the International Maritime Organization has organized an ad hoc working group for human elements issues. The task of the group is to prepare the measures against the negative effects of human elements on shipping and ocean environment. Human element has been the topic of the day in research field.The research is focused on these issues such as the mechanic of human error, the model of human error, the classification of human elements, the human behavior, human reliability analysis, measures for human error reduction. It really a great challenge for researchers in maritime area. Lot of difficulties has not been resolved up to now: lack of effective and reliable data of human elements, unclear of the relationship between the human errors and their caused factors, difficulty in identification of human elements involved in accidents, difficulty in confirm the results of human reliability analysis, and so forth. It is an insistent demand to overcome the difficulties.Take the ship collision avoidance as the research context, beginning with the modeling and classifying of human elements, the research will set the main objectives on overcoming the lack o f human elements, establishing relationship between unsafe acts and performance shaping factors, identifying the human elements in ship collision avoidance, constructing the models of human reliability analysis.The main contributions provided by the dissertation are listed as the following:1. A systematic review on current available human error models, human elements classifications, techniques of human elements identification, models of human reliability analysis have been carried out. Both of advantages and limitations of all those methods, approaches, classifications and models have been carefully analyses and listed in tables. The work provides a proper research platform for human reliability analysis.2. A human elements model and classification based on the human cognition during the ship collision avoidance has been established, which has provided a proper means for collecting human elements data and analyzing human errors. The model was constructed on the base of actional error models, with consideration of the demands of IMO and characteristic of ship collision. In the model, the recognition during collision avoidance has been divided into three phases: sensing, deciding and acting; the influencing factors on unsafe acts has been taken as performance shaping factors, which are further divided into four sub-classifications such as personal elements, organization,shipboard environment and natural environment. The model was designed with the hope that it could be used in some other kinds sea activities except for ship collision avoidance.3. Except questionnaires investigation and experiment of maritime simulator, a batch of written accident investigation reports has been used for human elements collecting, which has been proved to be an new effective approach of data collection. For the purpose of the research, 100 pieces of collision accident reports by 387 copies of questionnaires and 158 groups' data of simulator experiments have been collected, which provide an strong support for human reliability analysis.4. For the establishment of relationship between the unsafe acts and performance shaping factors, the technique of Data Mining has been used on the base of the data collecting from written accident reports. It seems that the approach is an effective and practicable ways in overcoming the difficulties in the filed. In the research, 6 unsafe acts and 9 p erformance s haping factors h ave b een i dentified. It c onfirmed that s ome n ew emerged mathematics techniques such as gray relational analysis; fuzzy logics are useful means in identifying human elements and influencing factors quantitatively.5. A new HRA model called THRASCA purposely designed for ship collision avoidance has been presented in the paper. The model is consisted of human error model, human classification and human reliability analysis model based on human recognition procedure during collision avoidance. With the model, the human reliability during collision avoidance could be assessed quantitatively. It could act as an aid for safety manager and ship handlers.6. The mechanics of human errors during collision avoidance has been researched with the THRASCA methods quantitatively. It shows that there is a nominal probability of collision valued around 0.0000554 to 0.0007, which are varied with the variation of performance shaping factors. The probability of human errors will be increased while the number of performance shaping factors increased. Hence, for collision avoidance, the measures should focus on the reduction of performance shaping factors.