We analyze complex maneuvering histories of ships obtained from training sessions on bridge simulators. The vessels carry out demanding operations at offshore oil installations, they have high maneuverability, and have Dynamical Positioning (DP) - an automatic maneuvering system using the propulsion devices and external position information. The maneuvering histories are recorded at high time resolution by the simulator software. In a specific simulator exercise for offshore crews, a ship is to be taken up to an installation consisting of three nearby platforms connected by bridges (Frigg field, North Sea), where a subsea inspection is to be carried out. When the position of operation is reached, an unmanned submarine (ROV) is launched. Somewhat later, all DP systems that yield external position references are disabled by the instructor. After this (unexpected) loss of external reference, most crews choose to maintain the vessel in position manually, without (full) support of the automation system (DP). As a result of different strategic choices, the maneuvering histories vary significantly between crews. We have, on one hand, developed measures for quantitative description of maneuvering histories, using vessel trace, velocity, and angle between hull and velocity. This applies to all three phases of the exercise: taking the vessel to position, keeping it in position under DP, and keeping it in position manually. On the other hand, we have obtained expert evaluations of the performance of each crew from experienced instructors. We compare the expert evaluations and the quantitative measures. We find a partial correlation: overall quality in task performance is reflected in more specific parameters.
展开▼