Issues on Nonlinear Programming for Multidisciplinary Design Optimization (MDO) in Ship Design Framework

摘要

This work is concerned with the study and the evaluation of formulations for Multidisciplinary Design Optimization (MDO) problems. The latter topic includes very difficult computational aspects, naturally arising from the simultaneous solution of several related optimization problems. In particular, in the naval applications deep interaction among different disciplines (i.e. optimization/feasibility problems) is claimed, in order to provide the solution. Each discipline concurs to give intermediate results to other disciplines, until the convergence of the overall framework is reached. The usual nonlinear programming techniques, including multiobjective optimization methods, seem still inadequate to cope with most of these challenging applications. In this work we first consider some relevant formulations proposed in the literature for MDO. Then, we consider the relationship between MDO and NonLinear Programming (NLP), with specific reference to feasibility and convergence analysis. Preliminary numerical results on a ship design application are included.