4.7 Article

Numerical instabilities in level set topology optimization with the extended finite element method

Journal

STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION
Volume 49, Issue 2, Pages 185-197

Publisher

SPRINGER
DOI: 10.1007/s00158-013-0982-x

Keywords

Topology optimization; Level sets; Extended finite element method; Enrichment strategy; Checker-boarding; Regularization; Perimeter constraint

Funding

  1. National Science Foundation [EFRI-SEED-1038305, CMMI-1201207]
  2. Directorate For Engineering
  3. Emerging Frontiers & Multidisciplinary Activities [1038305] Funding Source: National Science Foundation
  4. Div Of Civil, Mechanical, & Manufact Inn
  5. Directorate For Engineering [1201207] Funding Source: National Science Foundation

Ask authors/readers for more resources

This paper studies level set topology optimization of structures predicting the structural response by the eXtended Finite Element Method (XFEM). In contrast to Ersatz material approaches, the XFEM represents the geometry in the mechanical model by crisp boundaries. The traditional XFEM approach augments the approximation of the state variable fields with a fixed set of enrichment functions. For complex material layouts with small geometric features, this strategy may result in interpolation errors and non-physical coupling between disconnected material domains. These defects can lead to numerical instabilities in the optimized material layout, similar to checker-board patterns found in density methods. In this paper, a generalized Heaviside enrichment strategy is presented that adapts the set of enrichment functions to the material layout and consistently interpolates the state variable fields, bypassing the limitations of the traditional approach. This XFEM formulation is embedded into a level set topology optimization framework and studied with material-void and material-material design problems, optimizing the compliance via a mathematical programming method. The numerical results suggest that the generalized formulation of the XFEM resolves numerical instabilities, but regularization techniques are still required to control the optimized geometry. It is observed that constraining the perimeter effectively eliminates the emergence of small geometric features. In contrast, smoothing the level set field does not provide a reliable geometry control but mainly improves the convergence rate of the optimization process.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.7
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

No Data Available
No Data Available