Method for estimating parameters of practical ship manoeuvring models based on the combination of RANSE computations and System Identification
CHALLENGE - In this work a method for estimating parameters of practical ship maneuvering models based on the combination of RANSE computations and System Identification (SI) procedure is investigated, considering as test case a rather slender twin screw and two rudders ship.
SOLUTION - The SI process has been carried out coupling the modeFRONTIER genetic algorithm with the system based model. Before performing SI, a preliminary sensitivity analysis was conducted determining the most important coefficients in the maneuverability model considering both ZigZag and Turning Circle maneuvers. CFD simulations have been performed in order to provide the input data for the SI procedure. In this work three standard manoeuvres have been considered: the optimization procedure has been applied to two of them at the same time. Namely, the Turning Circle (ı = 35◦) and the ZigZag 10/10 have been considered since they have opposite characteristics, with high and low rudder angles and consequently high and low sway and yaw speeds.
BENEFITS - The genetic algorithm MOGA-II available in modeFRONTIER was used for this procedure. Overall the results show that the hybrid metric approach, with a combination of Hausdorff metric and Euclidean metric for trajectory and velocity variables is the best solution.