A decomposition method for exploiting parallel computing including the determination of an optimal number of subsystems

Sangjin Jung, Gyu Byung Park, Dong Hoon Choi

Research output: Contribution to journalArticlepeer-review

7 Scopus citations

Abstract

Many practical design problems are multidisciplinary and typically involve the transfer of complex information between analysis modules. In solving such problems, the method for performing multidisciplinary analyses greatly affects the speed of the total design time. Thus, it is very important to group and order a multidisciplinary analysis (MDA) process so as to minimize the total computational time and cost by decomposing a large multidisciplinary problem into several subsystems and then processing them in parallel. This study proposes a decomposition method that exploits parallel computing, including the determination of an optimal number of subsystems by using a multi-objective optimization formulation and a messy genetic algorithm (GA) modified to handle discrete design variables. In the suggested method, an MDA process is decomposed and sequenced for simultaneously minimizing the feedback couplings within each subsystem, the total couplings between subsystems, the variation of computation times among subsystems, and the computation time of each subsystem. The proposed method is applied to the decomposition of an artificial complex system example and a multidisciplinary design problem of a rotorcraft with 17 analysis modules; promising results are presented using this proposed method.

Original languageEnglish (US)
Article number041005
JournalJournal of Mechanical Design, Transactions Of the ASME
Volume135
Issue number4
DOIs
StatePublished - May 15 2013

All Science Journal Classification (ASJC) codes

  • Mechanics of Materials
  • Mechanical Engineering
  • Computer Science Applications
  • Computer Graphics and Computer-Aided Design

Fingerprint Dive into the research topics of 'A decomposition method for exploiting parallel computing including the determination of an optimal number of subsystems'. Together they form a unique fingerprint.

Cite this