Introduction of a clustering algorithm evaluated via the texture input of visco-plastic self-consistent (VPSC) computations

Deformation and process simulations including microstructural properties are known for their high complexity and time-demanding calculations. A challenging topic in this context is the local texture evolution during forming processes of industrially-sized parts. In this study, a meso-scale model, known as the visco-plastic self-consistent (VPSC) approach, is utilized to generate material texture and mechanical response. Although this model is known for its high simulation throughput, calculation times can increase drastically with finite element model size. The aim of this work is to investigate the effect of a reduced texture input on calculation time and mechanical response. For this purpose, a clustering algorithm, based on the k-means approach is proposed and applied to various textures. These are subsequently utilized as input for simulations of compression tests using Standalone VPSC and a VPSC coupled finite element model. A quantitative analysis of the results demonstrates a considerable reduction of calculation time by clustering without a significant loss of accuracy in mechanical material response. Additionally, a nearly identical trend in error values is observable across multiple tests and different texture intensities. • A general Clustering Algorithm is applied to reduce crystal orientations. • The performance is tested via VPSC and FE-VPSC coupled LSDYNA simulations. • An optimal clustering intensity (radius) was found by evaluation of flow curves. • Simulation effort/time was drastically reduced by using clustered texture.