A Contrastive Study of Two Heaviside-based Morphological Filters

Morphological filters can be used to eliminate the numerical instabilities in topology optimization results, including the checkerboard patterns, mesh-dependence and intermediate densities. In this paper, two morphological filters, namely Type 1 and Type 2 filters are constructed according to the morphology-based restriction schemes. Type 1 filters are constructed using the Heaviside filter and the improved Heaviside filter proposed by Sigmund and, Type 2 filters are constructed using the Heaviside filter and the improved Heaviside filter proposed by Guest. The optimization effects of the two kinds of morphological filters are investigated, including the convergence, the ability to obtain discrete designs, iteration times and compliances. Test results show that open and close filters exhibit better performances than the advanced filters, and the advanced filters are inclined to produce instabilities during the optimization process.