Scientific Visualization, 2019, volume 11, number 1, pages 107 - 118, DOI: 10.26583/sv.11.1.09
A Method for Topology Optimization Results Estimation Based on the Analysis of Raster Images of the Material Distribution
Authors: P.A. Alekhin1, A.O. Glebov2, S.V. Karpov3, S.V. Karpushkin4
Tambov State Technical University
1 ORCID: 0000-0002-0823-8369, pavel.aleh@yandex.ru
2 ORCID: 0000-0003-4214-6422, glebov.alex@inbox.ru
3 ORCID: 0000-0001-8238-1537, karpov.sv@mail.tstu.ru
4 ORCID: 0000-0001-6886-0646, karp@mail.gaps.tstu.ru
Abstract
Topology optimization is widespread technique in modeling the stress-strain state, thermal conductivity, electromagnetism, etc. As a rule, tasks are solving with SIMP method using the finite-element analysis. The efficiency of the SIMP method depends on the discretization of the computational domain into finite elements, the initial distribution of the material and the filtration procedures of the results obtained. In the previous papers about topology optimization the estimation of the calculated results is based only on the visual analysis of results material distribution. In this paper, we propose a method for estimation of results based on the analysis of raster images of the material distribution fields. The method involves calculation of ratio of pixels that characterize the presence of material and belong to the transition region by the total number of pixels of the calculation area. The method is universal, independent of the main calculation equations, and can be used for qualitative and quantitative comparison of the results of solving two-dimensional problems of topology optimization.
The influence of density of the calculated finite element mesh on the quality of the results of topology optimization of stationary heat conduction of a flat metal plate investigated. The method of moving asymptotes was chosen as an optimization method. The duration of calculations was examined; the optimal density value of a mesh was found and justified.
Keywords: topology optimization, image analysis, SIMP-method, finite element method, dendritic structure.
