Optimum design of laminated composite structures via a multilevel substructuring approach

Tai-Yan Kam*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

7 Scopus citations

Abstract

This paper presents a multilevel substructufing and optimization approach to the minimum weight design of laminated composite structures. The optimization process is carried out in a double scheme which consists of optimizations at system and subsystem levels. At the system level of optimization, an optimality criterion method is used to design component thicknesses which minimize structural weight subject to structural behavioral constraints as well as side constraints. At the subsystem level, the structure being divided into several substructures, fiber directions and layer thicknesses of each substructure are determined to minimize its weight subject to component behavioral constraints as well as side constraints. The objective at the subsystem level is accomplished by carrying out the minimization process again in a double scheme where the quasi-Newton method is used at the first sub-level of optimization for the optimal design of fiber directions and an optimality criterion method at the second sub-level for layer thickness design. The optimal solution is obtained by iterating between the different levels of optimization. Appropriate connectivity conditions for linking different levels of optimization are introduced to ensure convergence of solution. The feasibility and application of the present approach is illustrated by an example of the optimal design of a single-cell, three bay, cantilevered boxbeam.

Original languageEnglish
Pages (from-to)81-100
Number of pages20
JournalEngineering Optimization
Volume19
Issue number2
DOIs
StatePublished - 1 Apr 1992

Keywords

  • boxbeams
  • buckling
  • composites
  • dynamics
  • Laminates
  • multilevel optimization
  • structural optimization

Fingerprint Dive into the research topics of 'Optimum design of laminated composite structures via a multilevel substructuring approach'. Together they form a unique fingerprint.

Cite this