Parallel computation of three-dimensional crystal plasticity homogenization method

抄録

The mechanical behavior of hexagonal close packed (HCP) metals is drastically different from that of face-centered cubic (FCC) and body-centered cubic (BCC) metals. An HCP metal generally shows less ductility and stronger anisotropy than FCC and BCC metals, because of two main reasons, i.e., the strong anisotropy in the crystalline scale and deformation twinning. Therefore, a formability assessment of HCP meta l is strongly expected in the engineering field. To improve the formability of HCP metals, the development of a numerical procedure based on the multiscale modeling is expected. The local necking behavior is an important phenomenon to trigger ductile failure of metal materials. However, the necking behavior analysis using the crystal plasticity model, which is a material model describing the crystalline scale behaviors of metallic materials reasonably, requires a high computational costa especially in case of the three-dimensional analysis.

In this study, the computational efficiency of the analysis using a three-dimensional coupled macro-micro analysis based on the homogenization method is significantly improved by parallel computing using the Message Passing Interface (MPI). Using the presented framework, a large deformation analysis of metallic material, in which a strain localization occurs, is conducted, and the parallel computational efficiency in large deformation analysis with strain localization is investigated in detail.