Comparison of Genetic Algorithm and Gradient-Based Method Applied to Ship Shape Optimization

概要

論文の詳細を見る
Considerable progress has been made in developing computational fluid dynamics(CFD) for simulating turbulent flows around real-world configurations with higher reliability and accuracy. In addition, shape optimization methods coupling with computer aided design(CAD) and CFD, are being applied to fluid-dynamic design. These automatic optimization approaches are expected to be a powerful design tool. However, since CFD analysis requires lots of CPU time-consuming, performance of numerical optimizers must be carefully investigated. A preliminary objective of the present work is to compare and evaluate the performance of two optimization algorithms: 1) a gradient-based method, and 2) a genetic algorithm(GA). Based on CAD/CFD integration system, a prototype design system with the two optimization algorithms, is constructed. Simplified CAD and grid generation modules, and a flow solver which solves the Reynolds-averaged Navier-Stokes equations, are included in the system. The system is finally applied to minimize total resistance of a thin ship at non-zero Froude number. The capabilities and limitations of the two optimization algorithms, are discussed.
社団法人日本船舶海洋工学会の論文
2004-03-25