Design engineers community seems to be still skeptical on the usefulness of numerical opti- mization. Criticisms are often focused on the simplicity of the optimization problem solved (e.g. resistance minimization) when compared to the complexity of a real-life design problem, involving tens of different criteria. Indeed, much more challenging requirements are on the desk of the design team and solutions tend to become more complex by the hour, particularly when new features, missions, or capabilities are to be considered for the final design. In order to address complex requirements, the design team typically explores different solutions by means of systematic variations of the hull form, while in a second phase the results are analyzed in order to reach the objectives. A much more efficient way is to adopt numerical optimization techniques. The main features requested to the new design (objectives and constraints) are translated in a mathematical form and then passed to a numerical optimization framework, able to manage all the tools necessary for the solution of the problem. Hence, new problems can be easily faced, discovering the right direction for the design development. In this paper some original Global Optimization (GO) algorithms are illustrated: a Multistart Gradient Method (MsGM), an evolutionary algorithm (Particle Swarm Optimization, PSO) and a Lipschitzian method (Diagonal Rectangular Algorithm for Global Optimization, DRAGO) are introduced and applied to the shape optimization of a cruise ship for improving safety and comfort, comparing the numerical results.
Global Optimization for Safety and Comfort
Daniele Peri;
2005
Abstract
Design engineers community seems to be still skeptical on the usefulness of numerical opti- mization. Criticisms are often focused on the simplicity of the optimization problem solved (e.g. resistance minimization) when compared to the complexity of a real-life design problem, involving tens of different criteria. Indeed, much more challenging requirements are on the desk of the design team and solutions tend to become more complex by the hour, particularly when new features, missions, or capabilities are to be considered for the final design. In order to address complex requirements, the design team typically explores different solutions by means of systematic variations of the hull form, while in a second phase the results are analyzed in order to reach the objectives. A much more efficient way is to adopt numerical optimization techniques. The main features requested to the new design (objectives and constraints) are translated in a mathematical form and then passed to a numerical optimization framework, able to manage all the tools necessary for the solution of the problem. Hence, new problems can be easily faced, discovering the right direction for the design development. In this paper some original Global Optimization (GO) algorithms are illustrated: a Multistart Gradient Method (MsGM), an evolutionary algorithm (Particle Swarm Optimization, PSO) and a Lipschitzian method (Diagonal Rectangular Algorithm for Global Optimization, DRAGO) are introduced and applied to the shape optimization of a cruise ship for improving safety and comfort, comparing the numerical results.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.