Model reduction in topology optimisation analysing the inner structure of sensitivity matrices

Topology optimisation models usually contain a great number of design variables and correspondingly lead to large matrices (pseudo load matrix and sensitivity matrix) which appear in sensitivity analysis. We apply singular value decomposition (SVD) to these matrices to analyse their inner structure. Based on the obtained information, we perform model reduction by transformation of the design variables into a lower‐dimensional space. Numerical examples illustrate the advocated theoretical concept. Reasonable results are obtained, based on only a fraction of all design variables. (© 2010 Wiley‐VCH Verlag GmbH & Co. KGaA, Weinheim)