Abstract: A current problem in maritime education and training using a ship handling simulator is how to judge quantitatively and objectively the skill progress of trainees for collision avoidance maneuvers without depending on an instructor's experience and subjectivity. The skill of performing a collision avoidance maneuver can be judged from the following three points: (1) timely action, (2) suitable operation, and (3) satisfactory result. In the present paper, whether collision avoidance action was taken in a timely way is evaluated with TCPA; whether a suitable operation was performed is evaluated by changes of heading angle; and, whether the result is satisfactory result is evaluated by Environmental Stress Value (ES value) and Time to Collision (TTC) which can be judged from the viewpoint of reducing risk of collision and stranding. A collision avoidance maneuver in a narrow water area is taken up in the case study. A total of 60 repetitions of simulator experiments were conducted systematically, and the validity of these objective indices was examined.

　

 Education and training using a ship handling simulator (SHS) arc already practiced globally. However, there is an important problem outstanding in education and training using a SHS. That is, it is necessary to propose an objective evaluation method to judge the technical skill-improvement of a trainee in the process of SHS training, without depending on the experience of instructor subjectively.

　

 The author has applied an Environmental Stress model ^{(1) (2)} to solve the problem in the execution process of SHS training, and has introduced a methodology for judging the results of the skill improvement for a trainee from the viewpoint of absorbing ship handling difficulty imposed on a trainee ⁽³⁾.

　

 In this research, in addition to the evaluation from such a viewpoint, an objective risk evaluation index, which pays attention to the exclusion of latent risk in the process of collision avoidance manoeuvres, was newly introduced.

　

 As for education and training using a SHS, it is commonly aimed at improving skills for evasive actions against collision and stranding by an officer on watch. Although there are many education and training elements required of an officer on watch, these are roughly divided into two categories; one element is action required of an officer (mariners' attitude), and the other is a technical element (ship handling technique). The former corresponds to lookout, checking ship's position, handling instruments, information exchange, regulation compliance, voyage plan and an emergency measures, etc. The latter corresponds to collision avoidance manoeuvres using main engine and rudder, etc.

　

 Although the skill of collision avoidance manoeuvres covers the whole process including detection of another ship, determination of collision risk, selection of collision avoidance action, execution, and confirming of maneuvering result, we focus here on three points: (1) timing, (2) which collision avoidance manoeuvres were performed, and (3) the results of collision avoidance manoeuvres.

　

 We believe that a judgment on the skill by which collision avoidance manoeuvres were completed is made from three viewpoints of: (1) good timing, (2) suitable operation, and (3) good result.

　

 Here, judging whether collision avoidance manoeuvres were performed with good timing is evaluated with TCPA, whether suitable operation was taken is evaluated by measuring the changed angle of own ship's course. These are the indices used from so far. On the other hand, on whether or not the result was sufficient result, apart from using only DCPA, we introduced the Environmental Stress Value (ES value) and Time to Collision (TTC) as indices of evaluation.

　

 Introduction of these indices is intended to judge the whole process of a series of collision avoidance manoeuvres from the viewpoints of both difficulty and safety, rather than a judgment only from the DTPA between 2 ships as the final result. ES value is a numerical index indicating the ship handling difficulty imposed on a mariner due to geographical restrictions of a water area and/or traffic congestion ^{(1) (2)}. Fundamentally, it is thought that the ship handling difficulty is directly proportional to danger of an accident. Therefore, the potential danger, which remains in a series of navigation processes, can be measured by reading changes of ES value.

　

 The ES value is expressed with the range of 0-1000; the stress ranking is as: from O to 500 is negligible; from 500 to 750 is marginal; from 750 to 900 is critical, and 900-1000 is catastrophic. On the relation between this stress ranking and the acceptable level, an ES value of 750 or more is a state that is unacceptable for mariners.

　

 TTC is calculated from Potential Area of Water (PAW, predicting tracks when presupposing that the states of operation and the states of external forces continue unchanged)⁽⁴⁾. Based on this concept of PAW, the TTC is calculated for every fixed time section from the time when the predicted track collides with another ship or an obstacle. This TTC is a numerical index indicating the potential risk of each time section of navigation. It can be judged that the danger of an accident is high when the value of TTC is close to zero.

　

4.1 Execution of Experiments

　

 The SHS of Kobe University of Mercantile Marine (KUMM) was used for the experiment evaluating of trainee's skill improvement, and 3 scenarios were used in the experiment. In each scenario, own ship encounters other ships in a narrow water area, as shown in Fig.1. Each scenario uses a 3,500TEU container ship as own ship passing through the narrow water area at 12 knots and a course of <<000>>. It encounters other ships in succession.

　

 Environmental conditions are no wind, no current, and daytime with clear visibility. Subjects are four students of KUMM. Although they are qualified as Third Grade Maritime Officer (Navigation), they have no experience on board as officers. These students had experienced the ship handling with a SHS in the past, and were already familiar with the equipment of this simulator.

　

 In implementing the experiment, distance and direction of a target ship are read from radar or repeater compass, the necessity of collision avoidance manoeuvres are judged by the subject, and the steering order to avoid a target ship is given to the helmsman. In addition, the ARPA function of radar is not used and the trainee performs collision avoidance manoeuvres mainly by changing own ship's course and not using the main engine as much as possible.

　

　

　

　

　

 Before the experiment, they were given the following information: confirmation about how to treat radar (such as distance, direction, range, and fixed distance marker, etc.), introduction of the sea area with a chart, and presentation of pilot card of own ship. Experiments were repeated 3 times for different scenarios in one day, but the number of repetitions was not informed to avoid a psychological influence.

　

 Having used scenario 1, 2, and 3 as one set, 5 repetitions of that set were conducted for one student. We also laid out a suitable schedule for the experiment at appropriate time intervals: after the 1^st set of experiments was conducted, then after passing through 4 or 5 days the 2^nd set of experiments was conducted.

　

　

4.2 Evaluation of Skill Improvement

　

4.2.1 Skill Improvement with Start Timing of Collision Avoidance Manoeuvres and Changed Angle of Course

　

 Only through experience do beginners acknowledge that if any action to avoid collision is delayed for target ships, own ship approaches too close to another ship resulting in operational difficulty and risk of collision. They learn the necessity of commencing early action to avoid collision. Consequently, using the same scenario, it is considered to be a skill improvement for trainees to take more TCPA for other ships as the experiments proceed.

　

Fig. 2 shows the change of initial TCPA as the training proceeds. And it shows the total average for 60 trainings with the 4 students. TCPA taken appears to a rise to the right, i.e., trend that a larger TCPA is taken as the training proceeds, is shown.

　

　

 Fig. 3 shows the changes in heading angle's average for 60 trainings with the 4 students as the number of times of training increases. Although it seems that 30 to 40 degrees could be taken as an average for the changed angle of course, it is almost uniform, and the correlation with the number of times of training is not clearly found.