Please use this identifier to cite or link to this item:
Title: On hole nucleation in topology optimization using the level set methods
Authors: Wang, S.Y.
Lim, K.M. 
Khoo, B.C. 
Wang, M.Y.
Keywords: Level set methods
Topology opti-
Issue Date: 2007
Citation: Wang, S.Y.,Lim, K.M.,Khoo, B.C.,Wang, M.Y. (2007). On hole nucleation in topology optimization using the level set methods. CMES - Computer Modeling in Engineering and Sciences 21 (3) : 219-238. ScholarBank@NUS Repository.
Abstract: Hole nucleation is an important issue not yet fully addressed in structural topology optimization using the level set methods. In this paper, a consistent and robust nucleation method is proposed to overcome the inconsistencies in the existing implementations and to allow for smooth hole nucleation in the conventional shape derivatives-based level set methods to avoid getting stuck at a premature local optimum. The extension velocity field is constructed to be consistent with the mutual energy density and favorable for hole nucleation. A negative extension velocity driven nucleation mechanism is established due to the physically meaningful driving force. An extension velocity filtering approach is developed to allow for nucleation of new holes at the sites where the material is ineffectively used while the ill-posed topology optimization problem can be regularized. To overcome the numerical instabilities caused by the level set evolution, the gradients of the level set function are kept bounded using a rescaling-based reinitialization scheme based on a global representation technique without moving the free boundary. Inconsistencies with the regularization and reinitialization techniques are eliminated and smooth nucleation of new holes becomes possible. The level set-based topology optimization would become more accurate and efficient. The success of the present method is demonstrated with the classical examples in minimum compliance design. Copyright © 2007 Tech Science Press.
Source Title: CMES - Computer Modeling in Engineering and Sciences
ISSN: 15261492
Appears in Collections:Staff Publications

Show full item record
Files in This Item:
There are no files associated with this item.

Page view(s)

checked on Oct 13, 2019

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.