Investigation and improvement of sensitivity computation using the area-fraction weighted fixed grid FEM and structural optimization

Peter D. Dunning*, H. Alicia Kim, Glen Mullineux

*Corresponding author for this work

Research output: Contribution to journalArticle

39 Citations (Scopus)

Abstract

Boundary based structural optimization methods often employ a fixed grid FEM to compute sensitivities for efficiency and simplicity. A simple and popular fixed grid approach is to modify the stiffness of elements intersected by the boundary by an area-fraction weighting. However, poor sensitivities and numerical instabilities can occur when using this method. Sensitivity computation for a compliance objective is investigated and the results are used to develop a weighted least squares scheme to improve sensitivities computed by the area-fraction approach. This is implemented to gether with a numerically stable structural topology optimization using the level set method with no additional filtering or regularization. The performance of the proposed scheme is demonstrated by classic benchmark examples of topology optimization. (C) 2011 Elsevier B.V. All rights reserved.

Original languageEnglish
Pages (from-to)933-941
Number of pages9
JournalFinite Elements in Analysis and Design
Volume47
Issue number8
DOIs
Publication statusPublished - Aug 2011

Keywords

  • fixed grid
  • area-fraction weighting
  • sensitivity computation
  • least squares method
  • level set method
  • structural topology optimization
  • level-set method
  • topology optimization
  • shape
  • design

Cite this

Investigation and improvement of sensitivity computation using the area-fraction weighted fixed grid FEM and structural optimization. / Dunning, Peter D.; Kim, H. Alicia; Mullineux, Glen.

In: Finite Elements in Analysis and Design, Vol. 47, No. 8, 08.2011, p. 933-941.

Research output: Contribution to journalArticle

@article{3cd0a71e75c6408aa06534faa065e7a2,
title = "Investigation and improvement of sensitivity computation using the area-fraction weighted fixed grid FEM and structural optimization",
abstract = "Boundary based structural optimization methods often employ a fixed grid FEM to compute sensitivities for efficiency and simplicity. A simple and popular fixed grid approach is to modify the stiffness of elements intersected by the boundary by an area-fraction weighting. However, poor sensitivities and numerical instabilities can occur when using this method. Sensitivity computation for a compliance objective is investigated and the results are used to develop a weighted least squares scheme to improve sensitivities computed by the area-fraction approach. This is implemented to gether with a numerically stable structural topology optimization using the level set method with no additional filtering or regularization. The performance of the proposed scheme is demonstrated by classic benchmark examples of topology optimization. (C) 2011 Elsevier B.V. All rights reserved.",
keywords = "fixed grid, area-fraction weighting, sensitivity computation, least squares method, level set method, structural topology optimization, level-set method, topology optimization, shape, design",
author = "Dunning, {Peter D.} and Kim, {H. Alicia} and Glen Mullineux",
year = "2011",
month = "8",
doi = "10.1016/j.finel.2011.03.006",
language = "English",
volume = "47",
pages = "933--941",
journal = "Finite Elements in Analysis and Design",
issn = "0168-874X",
publisher = "Elsevier",
number = "8",

}

TY - JOUR

T1 - Investigation and improvement of sensitivity computation using the area-fraction weighted fixed grid FEM and structural optimization

AU - Dunning, Peter D.

AU - Kim, H. Alicia

AU - Mullineux, Glen

PY - 2011/8

Y1 - 2011/8

N2 - Boundary based structural optimization methods often employ a fixed grid FEM to compute sensitivities for efficiency and simplicity. A simple and popular fixed grid approach is to modify the stiffness of elements intersected by the boundary by an area-fraction weighting. However, poor sensitivities and numerical instabilities can occur when using this method. Sensitivity computation for a compliance objective is investigated and the results are used to develop a weighted least squares scheme to improve sensitivities computed by the area-fraction approach. This is implemented to gether with a numerically stable structural topology optimization using the level set method with no additional filtering or regularization. The performance of the proposed scheme is demonstrated by classic benchmark examples of topology optimization. (C) 2011 Elsevier B.V. All rights reserved.

AB - Boundary based structural optimization methods often employ a fixed grid FEM to compute sensitivities for efficiency and simplicity. A simple and popular fixed grid approach is to modify the stiffness of elements intersected by the boundary by an area-fraction weighting. However, poor sensitivities and numerical instabilities can occur when using this method. Sensitivity computation for a compliance objective is investigated and the results are used to develop a weighted least squares scheme to improve sensitivities computed by the area-fraction approach. This is implemented to gether with a numerically stable structural topology optimization using the level set method with no additional filtering or regularization. The performance of the proposed scheme is demonstrated by classic benchmark examples of topology optimization. (C) 2011 Elsevier B.V. All rights reserved.

KW - fixed grid

KW - area-fraction weighting

KW - sensitivity computation

KW - least squares method

KW - level set method

KW - structural topology optimization

KW - level-set method

KW - topology optimization

KW - shape

KW - design

U2 - 10.1016/j.finel.2011.03.006

DO - 10.1016/j.finel.2011.03.006

M3 - Article

VL - 47

SP - 933

EP - 941

JO - Finite Elements in Analysis and Design

JF - Finite Elements in Analysis and Design

SN - 0168-874X

IS - 8

ER -