Efficient fast Fourier transform-based numerical implementation to simulate large strain behavior of polycrystalline materials

Abstract

In this paper, a new full-field numerical framework is proposed to model large strain phenomena in polycrystals. The proposed framework is based on the elasto-viscoplastic (EVP) fast Fourier transform (FFT) formulation presented by Lebensohn et al. (2012) and the rate dependent crystal plasticity framework developed by Asaro and Needleman (1985). In this implementation, the full-field solutions of micromechanical fields are computed on a regular, voxelized representative volume element (RVE) in which either a single or multiple grid points represent a single grain. The Asaro and Needleman (1985) formulation coupled with a semi-explicit, forward gradient time-integration scheme (Peirce et al., 1983) is used to compute local stresses and the FFT-based method is used to find local strain fluctuations at each grid point. The proposed model is calibrated using experimental uniaxial tensile test results of aluminum alloy (AA) 5754 sheet and then used to predict texture evolution and stress-strain response for balanced biaxial tension and plane-strain tension along rolling (RD) and transverse (TD) directions. The predicted stress-strain and texture results show a good agreement with experimental measurements. The CPU time required by the proposed model is compared with the original EVP-FFT model for two separate cases and the proposed model showed significant improvement in computation time (approximately 100 times faster).