ABSTRACT
Optimization taking uncertainty into consideration is usually referred to as robust optimization. Using Evolutionary Algorithms (EA) or genetic algorithm to solve robust optimization problems has been proposed in literature, a detailed overview can be found in (Jin & Branke 2005). In this paper, a robust design of ship structure considering uncertainty in compartment length is examined. However, a major challenge faced in robust optimization using EA is that it is a computationally cost-prohibitive process. In dealing with the uncertainty space, the optimization models can soon become intractable even for a normal size design problem. Marine structure design problems are considered as large scale problems, therefore additional measures must be taken so that the EA can work efficiently to allow these early stage trade studies to be applied in marine structural design space. Approaches that can ease the computational
1 INTRODUCTION
It is widely accepted that engineering designs are confronted with uncertainty. The variability of uncertainty in design variables and parameters can degrade the performance of an engineering system. Therefore, quantifying the effect of uncertainty in the early stage of marine structural design is necessary in order to avoid costly modifications at later stages. However, decision making under uncertainty in the early design stage process is troubled by the absence of information about exact form and distribution of the uncertainty. An interval uncertainty measure is proposed to address this concern where the uncertainty is only defined by an interval of potential values-no probabilistic description of the uncertainty domain is required. This paper presents a method that can fully account for the interval-type uncertainty in the optimization process while remaining computationally efficient.
