Effect of 3D Representative Volume Element (RVE) Thickness on Stress and Strain Partitioning in Crystal Plasticity Simulations of Multi-Phase Materials

Abstract

Crystal plasticity simulations help to understand the local deformation behavior of multi-phase materials based on the microstructural attributes. The results of such simulations are mainly dependent on the Representative Volume Element (RVE) size and composition. The effect of RVE thickness on the changing global and local stress and strain is analyzed in this work for a test case of dual-phase steels in order to identify the minimal RVE thickness for obtaining consistent results. 100×100×100 voxel representative volume elements are constructed by varying grain size and random orientation distribution in DREAM-3D. The constructed RVEs are sliced in depth up to 1, 5, 10, 15, 20, 25, 30, 40, and 50 layers to construct different geometries with increasing thickness. Crystal plasticity model parameters for ferrite and martensite are taken from already published data and assigned to respective phases. Although the global stress/strain behavior of different RVEs is similar (<5 % divergence), the local stress/strain partitioning in RVEs with varying thickness and grain size shows a considerable variation when statistically compared. It is concluded that two-dimensional (2D) RVEs can be used for crystal plasticity simulations when global deformation behavior is of interest. Whereas, it is necessary to consider three-dimensional (3D) RVEs, which have a specific thickness and number of grains for determining stabilized and more accurate local deformation behavior. This estimation will help researchers in optimizing the computation time for accurate mesoscale simulations.

Effect of the 3rd Dimension within the Representative Volume Element (RVE) on Damage Initiation and Propagation during Full-Phase Numerical Simulations of Single and Multi-Phase Steels

“…It is observed that the damage initiation in the case of 2D-RVE occurs at a global strain of ≈ 6% earlier in comparison with damage initiation in the case of 3D-RVE. This trend is exciting and probably occurs due to the higher contrast of accumulated strain in the case of 2D, as reported by earlier researchers [17,23]. These observations help generalize that damage initiation in 2D-RVE occurs at ≈ 6% less global strain than the damage initiation in 3D-RVE.…”

“…The mesh density in all the generated RVEs is . In the previous study by Qayyum et al [ 23 ] it has already been established that the 3D RVEs with size comprising of grains behave isotropically when loaded in any direction. Due to an overall neutral orientation distribution function (ODF) selection during RVE generation, a complete random orientation distribution of the grains is observed.…”

“…For the numerical simulation modeling, RVEs are virtually constructed using the open-source tool Dream.3D 6.5.121 [ 31 ]. The constructed RVEs are sliced into geometries with 01-layer and 50-layers thicknesses according to the methodology published earlier [ 23 ]. These RVEs are used as input geometries for the numerical simulation models implemented separately in both cases.…”

“…The major are: (a) identification of the CP model parameters for each phase, and (b) selection of an appropriate representative volume element (RVE), which should be small enough for reduced computation costs and large enough to be inclusive of all the essential material microstructure features. In a few recent publications [ 21 , 22 , 23 ] the authors tried to address this issue and identified appropriate RVEs for single and multi-phase materials.…”

“…Several researchers in the recent past has contributed their valuable research results to address these challenges [ 29 , 30 ]. The optimal RVE size is crucial, and therefore in continuation of the previous publication [ 23 ], this study is being carried out in a systematic order to incorporate damage criteria on the previously identified optimal RVEs.…”