Resistance reduction of a military ship by variable-accuracy metamodel-based multidisciplinary robust design optimization

A method for simulation-based multidisciplinary robust design optimization (MRDO) affected by uncertainty is presented, based on variable-accuracy metamodelling. The approach encompasses a variable level of refinement of the design of experiments (DoE) used for the metamodel training, a variable accuracy for the uncertainty quantification (UQ), and a variable level of coupling between disciplines for the multidisciplinary analysis (MDA). The results of the present method are compared with a standard MRDO, used as a benchmark and solved by fully coupled MDA and fully accurate UQ, without metamodels. The hull-form optimization of the DTMB 5415 subject to stochastic speed is presented. A two-way steady coupled system is considered, based on hydrodynamics and rigid-body equation of motion. The objective function is the expected value of the total resistance, and the design variables pertain to the modification of the hull form. The effectiveness and the efficiency of the present method are evaluated in terms of optimal design performances and number of simulations required to achieve the optimal design.