Fast models for deformation texture development to be integrated in FE models of metal forming operationsBy Paul Van Houtte (KULeuven)
Co-authors: B. Van Bael (KULeuven)
Q. Xie (Oak Ridge National Lab, USA)
M. Seefeldt (KULeuven)
FE simulations of metal forming processes can predict the strain history of every material 'location', which is in fact a representative volume element or RVE at the macroscale. In this there is a two-way coupling: the strain history affects the development of the anisotropy of these RVEs due to texture, which in turn affects the prediction of the strain histories. We consider three length scales: the 'engineering length scale' (part to be formed), the 'macroscopic length' scale (polycrystal, which has a texture) and the 'mesoscopic length scale' (grain, which has a crystal orientation). The engineering FE model covers the transition from the engineering scale to the macro-scale whereas the polycrystal deformation model connects the 'macroscale' to the 'mesoscale'. There are models for each length-scale transition. The meso-macro model is called 'polycrystal deformation model.; a finite element model could also be used for this, resulting in a 'FE2-model', which is heavily demanding on computer resources. Less demanding would be a 'Crystal-Plasticity Fast Fourier Transformation' model 'CP-FFT. Both models treat the interaction between the grains of the polycrystalline material in a serious way, at the level of stress equilibrium and compatibility of displacements at local grain boundaries. Instead one can use much simpler 'statistical models' which model grain interactions only in a statistical way. Such models are however orders of magnitude faster, and may even be pretty accurate as well [1-3]. Some of these models and their results will be discussed as well as serious attempts to further increase their accuracy by improving the compatibility of displacements at the grain boundaries (still in a statistical way).
Ⓒ Photos:Toerisme Leuven