Computational system for multi disciplinary optimization at conceptual design stage

Abstract

Presented is a computational system for multidisciplinary design optimization (MDO) at the conceptual design stage. During this phase, hundreds of low-fidelity models such as equations and compiled code, and thousands of variables are used to describe a complex product such as aircraft. In this context the paper presents a novel computational approach associated with the complete MDO process. The first aspect of the proposed approach is the dynamic derivation of the optimal computational plan for each design study, given the designer's choice of independent variables. The second aspect is the effectiveness with which the trade-off landscape is obtained. This is crucial from an engineering point of view, since such information will be used for selecting a baseline design. The approach is demonstrated with an aircraft design test case consisting of 96 models and 120 variables.