We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
Fourier-Accelerated Nodal Solvers (FANS) for homogenization problems.
- Authors
Leuschner, Matthias; Fritzen, Felix
- Abstract
Fourier-based homogenization schemes are useful to analyze heterogeneous microstructures represented by 2D or 3D image data. These iterative schemes involve discrete periodic convolutions with global ansatz functions (mostly fundamental solutions). The convolutions are efficiently computed using the fast Fourier transform. FANS operates on nodal variables on regular grids and converges to finite element solutions. Compared to established Fourier-based methods, the number of convolutions is reduced by FANS. Additionally, fast iterations are possible by assembling the stiffness matrix. Due to the related memory requirement, the method is best suited for medium-sized problems. A comparative study involving established Fourier-based homogenization schemes is conducted for a thermal benchmark problem with a closed-form solution. Detailed technical and algorithmic descriptions are given for all methods considered in the comparison. Furthermore, many numerical examples focusing on convergence properties for both thermal and mechanical problems, including also plasticity, are presented.
- Subjects
ASYMPTOTIC homogenization; MICROSTRUCTURE; ITERATIVE methods (Mathematics); FAST Fourier transforms; FINITE element method
- Publication
Computational Mechanics, 2018, Vol 62, Issue 3, p359
- ISSN
0178-7675
- Publication type
Article
- DOI
10.1007/s00466-017-1501-5