A sequential decision process for the system-level design of structural frames

Mehmet Unal, Simon W. Miller, Jaskanwal P.S. Chhabra, Gordon P. Warn, Michael A. Yukish, Timothy W. Simpson

Research output: Contribution to journalArticlepeer-review

13 Scopus citations

Abstract

The building design community is currently experiencing a shift towards generating more resilient and sustainable designs that are also safe and economic. Integrating these broad, often conflicting, factors into design causes the design process to become more complex, the decisions more difficult, and a need for higher fidelity (and more expensive) modeling efforts to gain insight to resolve these tradeoffs. Set-based design is a promising alternative to traditional point-based design for such complex design problems because many design concepts are generated, refined and carefully eliminated throughout the process thereby maintaining significant freedom in the early stages of design. There is an emerging concept of closely coupling set-based design with model-based simulation to systematically contract the design space through a sequence of modeling efforts in which bounding models are constructed to guarantee the antecedent model only eliminates design alternatives that are dominated when analyzed using a subsequent higher fidelity model. In this paper, the concept of the sequential decision process is explored for the system-level design of seismic-resisting structural frames using nonlinear static analysis. Bounding models are derived for this design problem that are sufficiently general for various types of structural frames. The merit of the novel approach lies in giving designers the ability to search the tradespace exhaustively, arriving at the global minimum with a reduced cost by comparison to a full evaluation using the highest fidelity model while also providing the freedom to retain dominated and non-dominated solutions.

Original languageEnglish (US)
Pages (from-to)991-1011
Number of pages21
JournalStructural and Multidisciplinary Optimization
Volume56
Issue number5
DOIs
StatePublished - Nov 1 2017

All Science Journal Classification (ASJC) codes

  • Software
  • Control and Systems Engineering
  • Computer Science Applications
  • Computer Graphics and Computer-Aided Design
  • Control and Optimization

Fingerprint Dive into the research topics of 'A sequential decision process for the system-level design of structural frames'. Together they form a unique fingerprint.

Cite this