A novel approach for anisotropic hardening modeling. Part II: Anisotropic hardening in proportional and non-proportional loadings, application to initially isotropic material
Résumé
The modeling of anisotropic hardening, in particular for non-proportional loading paths, is a challenging task for advanced macroscopic models. The complex distortion of the yield locus is related to the activation and cross-hardening of different slip systems, depending on crystallographic orientations. These physical mechanisms can be taken into account in polycrystalline models but the computation times are enormous. The novel approach detailed in Part I (Rousselier et al., 2009) consists in: (i) drastically reducing the number of crystallographic orientations to save the computation cost, (ii) applying a parameter calibration procedure to obtain a good agreement with the experimental database. This methodology is first applied here to the anisotropic hardening in the proportional loadings of the strongly anisotropic aluminum alloy of Part I. Very good modeling is achieved with only eight crystallographic orientations. Different levels of additional hardening in biaxial proportional loading as compared to uniaxial loading can be modeled with the same polycrystalline model. For this, only the parameter calibration has to be performed with different databases. The same methodology has also been applied for the modeling of isotropic behavior. The best compromise between model accuracy and numerical cost is obtained with fourteen orientations. The deviations from isotropy are acceptable in all loading directions. Different levels of hardening in orthogonal loading: simple shear followed by simple tension, are achieved without any modification of the model equations. Only the parameter calibration has to be performed with different hardening levels in the database. FE calculations of a deep drawing test have been performed. The CPU time of the polycrystalline model is only five times larger than that with the simple von Mises model. The CPU time with texture evolution is further increased by a factor of two. The effects of texture evolution in rolling of the initially isotropic fcc material have been investigated. The resulting texture and hardening are qualitatively good.