Perpustakaan Digital ITB

This thesis discuss how the exibility of a variable-stiffness n panel model affects the generated net thrust. An optimisation algorithm, global surrogate-assisted genetic algorithm with Kriging was used to optimise the net thrust force produced by the n panel model, where kriging was used to construct a surrogate model to accelerate the optimisation process. Fitness value was evaluated by directly measuring the thrust force in towing tank experiment. Fin panel model was created using silicon rubber embedded with six spring wires. The n panel's exibility can be varied by changing the length of six springs embedded within the silicon rubber, which enables the n panel to have exibility range of 5.5 MPa to 200 GPa. The data shows that there exist an optimum stiffness condition which holds true for different kinematical parameters variation. However, different stiffness con gurations also change the optimum frequency of the n, with lower frequency n responds favorably to lower stiffness con guration n due to different bending response for a given material properties. It is also shown that there exist a more prominent kinematical parameter which produce the larger optimum thrust compared to the other, less sensitive, kinematical parameter.