High Fidelity Modelling and Parallel Architecture in the Simulation Based Design of a Fast Multihull Ship

Design activities of marine vehicles usually involve many different disciplines, highly interacting each other. In this context, design optimization meets great difficulties due to the requested effort. Fast, multihull ships are the today's trend for dis- placement marine vehicles. However, if the space between the hulls is small, interference effects arise, generated both by the interaction between the boundary layers of the two hulls and by the two wave systems interaction. This leads to an increase in the total resistance of the ship that has to be minimized. Furthermore, also the effect of the real pitch angle and draught the ship has once at speed must to be accounted for. As a consequence, expensive RANS solvers, with free-surface capturing capacities and also able to handle the free-running model, are mandatory, which implies a dramatic increase of the total cost of the optimization cycle: to this aim, a number of different techniques have been developed in the field of Multi-Objective Design Optimization in order to alleviate the computational cost of the whole optimization cycle. In this paper, a real-life optimization problem for the shape optimization of a fast catamaran ship is tackled. Parallel architectures, in conjunction with metamodel techniques are applied. A mixed strategy, with the use of two different derivative-free global optimization algorithms, is presented. Preliminary results demonstrate the great improvements the designer may obtain by controlling the different elements of the interference between the hulls.