Reliable design activities of marine vehicles usually involve expensive solvers, dealing with the free-surface ow of a free running ship. This implies a huge cost for the full optimization cycle. To reduce it, dierent techniques have been developed limiting the number of calls to high-delity solvers. Variable Fidelity Modeling (VFM) is one of them: it manages two dierent level of delity, taking advantage from the good features of both. Up to now, this technique has been applied to local optimization algorithms only. In this paper, some test case of the shape optimization of ships are tackled, where a parallel VFM is applied. The novelty is the use of derivative-free global optimization algorithms. Preliminary results demonstrate the great time reductions that may be obtained by this technique.
Variable Fidelity Modeling and Parallel Architecture in the Global Optimization of a Ship
2008
Abstract
Reliable design activities of marine vehicles usually involve expensive solvers, dealing with the free-surface ow of a free running ship. This implies a huge cost for the full optimization cycle. To reduce it, dierent techniques have been developed limiting the number of calls to high-delity solvers. Variable Fidelity Modeling (VFM) is one of them: it manages two dierent level of delity, taking advantage from the good features of both. Up to now, this technique has been applied to local optimization algorithms only. In this paper, some test case of the shape optimization of ships are tackled, where a parallel VFM is applied. The novelty is the use of derivative-free global optimization algorithms. Preliminary results demonstrate the great time reductions that may be obtained by this technique.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
