High-fidelity Models in Simulation Based Design

This work presents a Simulation Based Design en- vironment based on a Global Optimization (GO) al- gorithm for the solution of optimum ship design problems. The procedure, illustrated in the frame- work of multiobjective optimization problems, make use of high-fidelity, CPU time expensive computa- tional models, including a free surface capturing RANSE solver. The use of GO prevents the optimizer to be trapped into local minima. The optimization is composed by global and local phases. In the global stage of the search, a few com- putationally expensive simulations are needed for creating surrogate models (metamodels) of the objec- tive functions. Tentative design, created to explore the design space are evaluated with these inexpensive analytical approximations. The more promising de- signs are clustered, then locally minimised and even- tually verified with high-fidelity simulations. New exact values are used to improve the metamodels and repeated cycles of the algorithm are performed. A Decision Maker strategy is finally adopted to select the more promising design. Starting for this solution, the final local refinement stage is performed by a gradient based local optimization technique. A key point in the algorithm is the reduction of the overall time needed for the objective functions evaluation. As an alternative to adjoint formulations an attractive technique is adopted, the Approximation Management Framework (AMF), based on a com- bined variable fidelity model - trust region strategy. Numerical examples are given demonstrating both the validity and usefulness of the proposed GO ap- proach.