Multi-objective surrogate based hull-form optimization using high-fidelity rans computations

Abstract

RANS-based optimization procedures for ship design become increasingly complex

and require the development of more efficient optimization techniques. The four phases of the design procedure are: shape parameterization, global sensitivity analysis, multi-objective optimization and design review. The dimensions of the design space can be mitigated by a smart choice for the shape parameterization and by screening and ranking the design variables in the global sensitivity phase. Subsequently, Surrogate Based Global Optimization (SBGO) is used to reduce the cost of the multi-objective optimization phase. For a practical application it is shown that the computational time reduces from two weeks to only a day when using SBGO instead of applying a Multi-Objective Genetic Algorithm (MOGA) directly to the solver. The design review phase is then used to verify and further develop the optimal design. Here, we focus on automatic ship design techniques which comprises the first three steps of the design procedure. Accelerating the ship design process is subject of ongoing research at the Maritime Research Institute Netherlands, making it useful for practical applications with turnaround times

of only a few weeks.