ANN-aided evaluation of dual-phase microstructural fabric tensors for continuum plasticity representation

Abstract

The evolution and orientation-dependent behavior of microstructures in dual-phase materials significantly affect the mechanical properties. How to quantify the microstructural effect in a continuum constitutive model, especially considering anisotropic elastic–plastic properties, is still a tough research topic for multiscale mechanics. In the present paper, the fabric tensors are successfully correlated with elastic–plastic mechanical properties of dual-phase materials with the help of the artificial neural network (ANN). The fabric tensors can be decomposed into isotropic and deviatoric components, which describe voluminal changes and orientation-dependent properties of the microstructural material through data-driven analysis. A correlation analysis combined with gradient-based attributions revealed, furthermore, that a lower-order fabric tensor with fewer components was sufficient for the complex morphology of microstructures. For crystal symmetric materials, the second-order fabric tensors are sufficient to generate an adequate description of anisotropic dual-phase microstructures. The fabric tensors provide a bridge to connect the microstructural characteristics with phenomenological continuum plasticity for complex materials. • Verified the correlation between fabric tensors and elastic–plastic mechanical properties. • Achieved an adequate description of dual-phase microstructures by second-order fabric tensors. • Identified the key fabric tensor components through data-driven analysis. • Provided a microstructural material’s state variable for continuum plasticity.