Optimal design of composite shells based on minimum weight and maximum feasibility robustness

• Type: Article in International Scientific Journal
• Date: 2017

Authors

Carlos Alberto Conceição António (Author) FEUP; Hoffbauer, LN (Author) Other

Journal

Title: International Journal of Mechanics and Materials in Design

Vol. 13 No. 2

Pages: 287-310

ISSN: 1569-1713

Publisher: Springer Nature

Indexing

ISI Web of Knowledge - 9 Citations

Scopus - 10 Citations

Other information

Authenticus ID: P-00K-14C

DOI: 10.1007/s10999-015-9329-7

Abstract (EN): A robust design optimization (RDO) approach for minimum weight and safe shell composite structures with minimal variability into design constraints under uncertainties is proposed. A new concept of feasibility robustness associated to the variability of design constraints is considered. So, the feasibility robustness is defined through the determinant of variance-covariance matrix of constraint functions introducing in this way the joint effects of the uncertainty propagations on structural response. A new framework considering aleatory uncertainty into RDO of composite structures is proposed. So, three classes of variables and parameters are identified: deterministic design variables, random design variables and random parameters. The bi-objective optimization search is performed using on a new approach based on two levels of dominance denoted by Co-Dominance-based Genetic Algorithm (CoDGA). The use of evolutionary concepts together sensitivity analysis based on adjoint variable method is a new proposal. The examples with different sources of uncertainty show that the Pareto front definition depends on random design variables and/or random parameters considered in RDO. Furthermore, the importance to control the uncertainties on the feasibility of constraints is demonstrated. CoDGA approach is a powerfully tool to help designers to make decision establishing the priorities between performance and robustness.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 24