Computer Science · Computational Theory and Mathematics
Ship design optimization considering uncertainty
See on the mapAbstract
Multi-disciplinary analysis and design optimization allow designers to break away from the traditional design spiral approach and focus on searching the design space for the best overall design rather than considering best discipline-specific designs. The effects of uncertainty in the disciplinary calculations of complex design simulation models are considered. Uncertainty is quantified using a of calculation that is developed based on the Mean Value Method. The confidence of success is the probability that a design will satisfy all constraints and meet performance objectives. Multiple sources of uncertainty are quantified and represented in a ship synthesis model. Optimizations using confidence of success as a third objective, in addition to cost and effectiveness, in a multi-objective genetic algorithm are executed. It is demonstrated that designs lying along constraint boundaries, which are typically identified as optimum designs in deterministic design optimizations, likely have a very low chance of being successful.