Optimization of crashworthy marine structures

History shows that ferry and RoPax collisions with tankers can be devastating for human life. This paper follows up such a scenario to contribute to rational increase of safety of marine structures. Through the coupling of multi-objective structural optimization and crashworthiness analysis, a conventional tanker structure is optimized for higher collision tolerance, accounting for the change in hull mass, so that the increase in safety is efficient. Two new concepts, proposed here, are deemed necessary for the successful execution of this task: a ‘two-stage’ optimization approach, reducing the number of needed collision simulations, and a rapid collision simulation approach that utilizes coarse FE mesh and reduces calculation time. Combining the obtained results with the state-of-the-art knowledge, a new insight about crashworthy design of tanker structures is also realized.     

Vectorization and constraint grouping to enhance optimization of marine structures

Vectorization converts classical scalar optimization formulation, which strictly separates the objective from constraints, into a vector-based optimization, transforming constraints into objectives. Effectively, the search is not conducted anymore for a single optimum, but for a set of Pareto optima between the original objective and transformed constraints. Constraint grouping enhances handling of multiple constraints for vectorized problems, by combining several constraints within a single-objective function, thus reducing the computational time and computational difficulties of high-dimensional spaces created by vectorization. This paper formulates and investigates these two concepts with respect to design of marine structures. It analyses their effects on the possibility to improve the flexibility of optimization in a practical environment, by implementing them within a simple genetic algorithm. Obtained results of vectorization applied to realistic weight optimization problem are encouraging when compared with the results of the classical scalar form optimization, showing a significant improvement in magnitude as well as in reduced computation time needed to reach the optimum.